path: root/doc/autogen-texi.txt

@c -*- Mode: texinfo -*-
@setfilename autogen.info
@ignore
This file serves two purposes:

1) it provides that stupid (at)setfilename so that automake will
   deign to produce the documentation

2) a text repository for documentation that would make the doc
   template more confusing.


  This file is part of AutoGen.
  AutoGen Copyright (c) 1992-2012 by Bruce Korb - all rights reserved

  AutoGen is free software: you can redistribute it and/or modify it
  under the terms of the GNU General Public License as published by the
  Free Software Foundation, either version 3 of the License, or
  (at your option) any later version.

  AutoGen is distributed in the hope that it will be useful, but
  WITHOUT ANY WARRANTY; without even the implied warranty of
  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  See the GNU General Public License for more details.

  You should have received a copy of the GNU General Public License along
  with this program.  If not, see <http://www.gnu.org/licenses/>.
@end ignore

@ignore
START == COMMENTS == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
Resume input from autogen.texi
@end ignore

@c === SECTION MARKER

@node Predefines
@section Pre-defined Names
@cindex predefines

When AutoGen starts, it tries to determine several names from the
operating environment and put them into environment variables for use in
both @code{#ifdef} tests in the definitions files and in shell scripts
with environment variable tests.  @code{__autogen__} is always defined.
For other names, AutoGen will first try to use the POSIX version of the
@code{sysinfo(2)} system call.  Failing that, it will try for the POSIX
@code{uname(2)} call.  If neither is available, then only
"@code{__autogen__}" will be inserted into the environment.
In all cases, the associated names are converted to lower case, surrounded
by doubled underscores and non-symbol characters are replaced with
underscores.

With Solaris on a sparc platform, @code{sysinfo(2)} is available.
The following strings are used:

@itemize @bullet
@item
@code{SI_SYSNAME} (e.g., "__sunos__")
@item
@code{SI_HOSTNAME} (e.g., "__ellen__")
@item
@code{SI_ARCHITECTURE} (e.g., "__sparc__")
@item
@code{SI_HW_PROVIDER} (e.g., "__sun_microsystems__")
@item
@code{SI_PLATFORM} (e.g., "__sun_ultra_5_10__")
@item
@code{SI_MACHINE} (e.g., "__sun4u__")
@end itemize

For Linux and other operating systems that only support the
@code{uname(2)} call, AutoGen will use these values:

@itemize @bullet
@item
@code{sysname} (e.g., "__linux__")
@item
@code{machine} (e.g., "__i586__")
@item
@code{nodename} (e.g., "__bach__")
@end itemize

By testing these pre-defines in my definitions, you can select
pieces of the definitions without resorting to writing shell
scripts that parse the output of @code{uname(1)}.  You can also
segregate real C code from autogen definitions by testing for
"@code{__autogen__}".

@example
#ifdef __bach__
  location = home;
#else
  location = work;
#endif
@end example

@c === SECTION MARKER

@node Comments
@section Commenting Your Definitions
@cindex comments

The definitions file may contain C and C++ style comments.

@example
/*
 *  This is a comment.  It continues for several lines and closes
 *  when the characters '*' and '/' appear together.
 */
// this comment is a single line comment
@end example

@c === SECTION MARKER

@node Example
@section What it all looks like.

@noindent
This is an extended example:

@example
autogen definitions @samp{template-name};
/*
 *  This is a comment that describes what these
 *  definitions are all about.
 */
global = "value for a global text definition.";

/*
 *  Include a standard set of definitions
 */
#include standards.def

a_block = @{
    a_field;
    a_subblock = @{
        sub_name  = first;
        sub_field = "sub value.";
    @};

#ifdef FEATURE
    a_subblock = @{
        sub_name  = second;
    @};
#endif

@};
@end example

@ignore
END   == COMMENTS == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == TEMPLATE == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c === SECTION MARKER

@node Alternate Definition
@section Alternate Definition Forms
@cindex Alternate Definition

There are several methods for supplying data values for templates.

@table @samp
@item no definitions
It is entirely possible to write a template that does not depend upon
external definitions.  Such a template would likely have an unvarying
output, but be convenient nonetheless because of an external library
of either AutoGen or Scheme functions, or both.  This can be accommodated
by providing the @code{--override-tpl} and @code{--no-definitions}
options on the command line.  @xref{autogen Invocation}.

@item CGI
AutoGen behaves as a CGI server if the definitions input is from stdin
and the environment variable @code{REQUEST_METHOD} is defined
and set to either "GET" or "POST", @xref{AutoGen CGI}.  Obviously,
all the values are constrained to strings because there is no way
to represent nested values.

@item XML
AutoGen comes with a program named, @code{xml2ag}.  Its output can
either be redirected to a file for later use, or the program can
be used as an AutoGen wrapper.  @xref{xml2ag Invocation}.

The introductory template example (@pxref{Example Usage}) can be rewritten
in XML as follows:

@example
<EXAMPLE  template="list.tpl">
<LIST list_element="alpha"
      list_info="some alpha stuff"/>
<LIST list_info="more beta stuff"
      list_element="beta"/>
<LIST list_element="omega"
      list_info="final omega stuff"/>
</EXAMPLE>
@end example

A more XML-normal form might look like this:
@example
<EXAMPLE  template="list.tpl">
<LIST list_element="alpha">some alpha stuff</LIST>
<LIST list_element="beta" >more beta stuff</LIST>
<LIST list_element="omega">final omega stuff</LIST>
</EXAMPLE>
@end example
@noindent
but you would have to change the template @code{list_info} references
into @code{text} references.

@item standard AutoGen definitions
Of course.  :-)

@end table

@ignore
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
@end ignore
@page
@node Template File
@chapter Template File
@cindex template file
@cindex .tpl file

The AutoGen template file defines the content of the output text.
It is composed of two parts.  The first part consists of a pseudo
macro invocation and commentary.  It is followed by the template proper.

@cindex pseudo macro
@cindex macro, pseudo
This pseudo macro is special.  It is used to identify the file as a
AutoGen template file, fixing the starting and ending marks for
the macro invocations in the rest of the file, specifying the list
of suffixes to be generated by the template and, optionally, the
shell to use for processing shell commands embedded in the template.

AutoGen-ing a file consists of copying text from the template to the
output file until a start macro marker is found.  The text from the
start marker to the end marker constitutes the macro text.  AutoGen
macros may cause sections of the template to be skipped or processed
several times.  The process continues until the end of the template is
reached.  The process is repeated once for each suffix specified in the
pseudo macro.

This chapter describes the format of the AutoGen template macros
and the usage of the AutoGen native macros.  Users may augment
these by defining their own macros, @xref{DEFINE}.

@menu
* pseudo macro::       Format of the Pseudo Macro
* naming values::      Naming a value
* expression syntax::  Macro Expression Syntax
* AutoGen Functions::  AutoGen Scheme Functions
* Common Functions::   Common Scheme Functions
* native macros::      AutoGen Native Macros
* output controls::    Redirecting Output
@end menu

@c === SECTION MARKER

@node pseudo macro
@section Format of the Pseudo Macro
@cindex pseudo macro

The pseudo macro is used to tell AutoGen how to process a template.
It tells autogen:

@enumerate
@item
The start macro marker.  It consists of punctuation characters used to
demarcate the start of a macro.  It may be up to seven characters long and
must be the first non-whitespace characters in the file.

@noindent
It is generally a good idea to use some sort of opening
bracket in the starting macro and closing bracket in the ending
macro  (e.g. @code{@{}, @code{(}, @code{[}, or even @code{<}
in the starting macro).  It helps both visually and with editors
capable of finding a balancing parenthesis.

@item
That start marker must be immediately followed by the identifier strings
"AutoGen5" and then "template", though capitalization is not important.
@end enumerate

@noindent
The next several components may be intermingled:

@enumerate 3
@item
Zero, one or more suffix specifications tell AutoGen how many times to
process the template file.  No suffix specifications mean that it is to
be processed once and that the generated text is to be written to
@file{stdout}.  The current suffix for each pass can be determined with the
@code{(suffix)} scheme function (@pxref{SCM suffix}).

The suffix specification consists of a sequence of POSIX compliant file name
characters and, optionally, an equal sign and a file name formatting
specification.  That specification may be either an ordinary sequence of
file name characters with zero, one or two "%s" formatting sequences in it,
or else it may be a Scheme expression that, when evaluated, produces such a
string.  The Scheme result may not be empty.  The two string arguments
allowed for that string are the base name of the definition file, and the
current suffix (that being the text to the left of the equal sign).  (Note:
"POSIX compliant file name characters" consist of alphanumerics plus the
period (@code{.}), hyphen (@code{-}) and underscore (@code{_}) characters.)

If the suffix begins with one of these three latter characters and
a formatting string is not specified, then that character is presumed to
be the suffix separator.  Otherwise, without a specified format string,
a single period will separate the suffix from the base name in constructing
the output file name.

@item
Shell specification: to specify that the template was written expecting a
particular shell to run the shell commands.  By default, the shell used is the
autoconf-ed @code{CONFIG_SHELL}.  This will usually be @file{/bin/sh}.  The
shell is specified by a hash mark (@code{#}) followed by an exclamation mark
(@code{!}) followed by a full-path file name (e.g. @file{/usr/xpg4/bin/sh} on
Solaris):
@example
[= Autogen5 Template c
#!/usr/xpg4/bin/sh
=]
@end example

@item
Comments: blank lines, lines starting with a hash mark (@code{#}) and not
specifying a shell, and edit mode markers (text between pairs of @code{-*-}
strings) are all treated as comments.

@item
Some scheme expressions may be inserted in order to make configuration
changes before template processing begins.  ``@i{before template
processing begins}'' means that there is no current output file, no current
suffix and, basically, none of the AutoGen specific functions
(@pxref{AutoGen Functions}) may be invoked.

The scheme expression can also be used, for example, to save a pre-existing
output file for later text extraction (@pxref{SCM extract}).

@example
(shellf "mv -f %1$s.c %1$s.sav" (base-name))
@end example
@end enumerate

@noindent
After these must come the end macro marker:

@enumerate 6
@item
The punctuation characters used to demarcate the end of a macro.
Like the start marker, it must consist of seven or fewer punctuation
characters.
@end enumerate

The ending macro marker has a few constraints on its content.  Some of
them are just advisory, though.  There is no special check for advisory
restrictions.

@itemize @bullet
@item
It must not begin with a POSIX file name character (hyphen @code{-},
underscore @code{_} or period @code{.}), the backslash (@code{\}) or
open parenthesis (@code{(}).  These are used to identify a suffix
specification, indicate Scheme code and trim white space.

@item
If it begins with an equal sign, then it
must be separated from any suffix specification by white space.

@item
The closing marker may not begin with an open parenthesis, as that is used
to enclose a scheme expression.

@item
It cannot begin with a backslash, as that is used to indicate white
space trimming after the end macro mark.  If, in the body of the template,
you put the backslash character (@code{\}) before the end macro mark, then
any white space characters after the mark and through the newline character
are trimmed.

@item
It is also helpful to avoid using the comment marker (@code{#}).
It might be seen as a comment within the pseudo macro.

@item
You should avoid using any of the quote characters@:  double,
single or back-quote.  It won't confuse AutoGen, but it might well
confuse you and/or your editor.
@end itemize

As an example, assume we want to use @code{[+} and @code{+]} as the start
and end macro markers, and we wish to produce a @file{.c} and a @file{.h}
file, then the pseudo macro might look something like this:

@example
[+ AutoGen5 template -*- Mode: emacs-mode-of-choice -*-
h=chk-%s.h
c
# make sure we don't use csh:
(setenv "SHELL" "/bin/sh")  +]
@end example

The template proper starts after the pseudo-macro.  The starting
character is either the first non-whitespace character or the first
character after the newline that follows the end macro marker.

@c === SECTION MARKER

@node naming values
@section Naming a value
@cindex naming values

When an AutoGen value is specified in a template, it is specified by name.
The name may be a simple name, or a compound name of several components.
Since each named value in AutoGen is implicitly an array of one or more
values, each component may have an index associated with it.

@noindent
It looks like this:

@example
comp-name-1 . comp-name-2 [ 2 ]
@end example

Note that if there are multiple components to a name, each component
name is separated by a dot (@code{.}).  Indexes follow a component name,
enclosed in square brackets (@code{[} and @code{]}).  The index may be
either an integer or an integer-valued define name.  The first component
of the name is searched for in the current definition level.  If not
found, higher levels will be searched until either a value is found,
or there are no more definition levels.  Subsequent components of the
name must be found within the context of the newly-current definition
level.  Also, if the named value is prefixed by a dot (@code{.}),
@cindex .
then the value search is started in the current context only.
Backtracking
@cindex backtrack
into other definition levels is prevented.

If someone rewrites this, I'll incorporate it.  :-)

@c === SECTION MARKER

@node expression syntax
@section Macro Expression Syntax
@cindex expression syntax

AutoGen has two types of expressions:  full expressions and basic ones.
A full AutoGen expression can appear by itself, or as the argument
to certain AutoGen built-in macros:  CASE, IF, ELIF, INCLUDE,
INVOKE (explicit invocation, @pxref{INVOKE}), and WHILE.
If it appears by itself, the result is inserted into the output.
If it is an argument to one of these macros, the macro code
will act on it sensibly.

You are constrained to basic expressions only when passing
arguments to user defined macros, @xref{DEFINE}.

The syntax of a full AutoGen expression is:

@example
[[ <apply-code> ] <value-name> ] [ <basic-expr-1> [ <basic-expr-2> ]]
@end example

How the expression is evaluated depends upon the presence or absence
of the apply code and value name.  The "value name" is the name of
an AutoGen defined value, or not.  If it does not name such a value,
the expression result is generally the empty string.  All expressions
must contain either a @code{value-name} or a @code{basic-expr}.

@menu
* apply code::           Apply Code
* basic expression::     Basic Expression
@end menu

@node apply code
@subsection Apply Code

The "apply code" selected determines the method of evaluating the
expression.  There are five apply codes, including the non-use
of an apply code.

@table @samp
@item no apply code
This is the most common expression type.
Expressions of this sort come in three flavors:

@table @samp
@item <value-name>
The result is the value of @code{value-name}, if defined.
Otherwise it is the empty string.

@item <basic-expr>
The result of the basic expression is the result of the full expression,
@xref{basic expression}.

@item <value-name> <basic-expr>
If there is a defined value for @code{value-name}, then the @code{basic-expr}
is evaluated.  Otherwise, the result is the empty string.
@end table

@item % <value-name> <basic-expr>
If @code{value-name} is defined, use @code{basic-expr} as a format
string for sprintf.  Then, if the @code{basic-expr} is either a back-quoted
string or a parenthesized expression, then hand the result to the
appropriate interpreter for further evaluation.  Otherwise, for single
and double quote strings, the result is the result of the sprintf operation.
Naturally, if @code{value-name} is not defined, the result is the empty
string.

For example, assume that @code{fumble} had the string value, @code{stumble}:
@example
[+ % fumble `printf '%%x\\n' $%s` +]
@end example
This would cause the shell to evaluate "@code{printf '%x\n' $stumble}".
Assuming that the shell variable @code{stumble} had a numeric value,
the expression result would be that number, in hex.  Note the need
for doubled percent characters and backslashes.

@item ? <value-name> <basic-expr-1> <basic-expr-2>
Two @code{basic-expr}-s are required.  If the @code{value-name} is
defined, then the first @code{basic-expr-1} is evaluated, otherwise
@code{basic-expr-2} is.

@item - <value-name> <basic-expr>
Evaluate @code{basic-expr} only if @code{value-name} is @i{not} defined.

@item ?% <value-name> <basic-expr-1> <basic-expr-2>
This combines the functions of @samp{?} and @samp{%}.  If @code{value-name} is
defined, it behaves exactly like @samp{%}, above, using @code{basic-expr-1}.
If not defined, then @code{basic-expr-2} is evaluated.

For example, assume again that @code{fumble} had the string value, @code{stumble}:
@example
[+ ?% fumble `cat $%s` `pwd` +]
@end example
This would cause the shell to evaluate "@code{cat $stumble}".
If @code{fumble} were not defined, then the result would be the name
of our current directory.
@end table

@node basic expression
@subsection Basic Expression

A basic expression can have one of the following forms:

@table @samp
@item 'STRING'
A single quoted string.  Backslashes can be used to protect single
quotes (@code{'}), hash characters (@code{#}), or backslashes (@code{\})
in the string.  All other characters of STRING are output as-is when the
single quoted string is evaluated.  Backslashes are processed before the hash
character for consistency with the definition syntax.  It is needed there
to avoid preprocessing conflicts.

@item "STRING"
A double quoted string.  This is a cooked text string as in C,
except that they are not concatenated with adjacent strings.
Evaluating "@code{STRING}" will output STRING with all
backslash sequences interpreted.

@item `STRING`
A back quoted string.  When this expression is evaluated, STRING
is first interpreted as a cooked string (as in `"STRING"') and
evaluated as a shell expression by the AutoGen server shell.  This
expression is replaced by the @file{stdout} output of
the shell.

@item (STRING)
A parenthesized expression.  It will be passed to the Guile
interpreter for evaluation and replaced by the resulting value.
If there is a Scheme error in this expression, Guile 1.4 and Guile 1.6
will report the template line number where the error occurs.  Guile 1.7
has lost this capability.

Guile has the capability of creating and manipulating variables that
can be referenced later on in the template processing.  If you define
such a variable, it is invisible to AutoGen.  To reference its value,
you must use a Guile expression.  For example,
@example
[+ (define my-var "some-string-value") +]
@end example
can have that string inserted later, but only as in:
@example
[+ (. my-var) +]
@end example

Additionally, other than in the @code{%} and @code{?%} expressions, the
Guile expressions may be introduced with the Guile comment character
(@code{;}) and you may put a series of Guile expressions within a single
macro.  They will be implicitly evaluated as if they were arguments
to the @code{(begin ...)} expression.  The result will be the
result of the last Guile expression evaluated.
@end table

@ignore
END   == TEMPLATE == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == MACROS == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c === SECTION MARKER

@node native macros
@section AutoGen Native Macros
@cindex native macros

This section describes the various AutoGen natively defined macros.
Unlike the Scheme functions, some of these macros are "block macros"
with a scope that extends through a terminating macro.  Block macros
must not overlap.  That is to say, a block macro started within the
scope of an encompassing block macro must have its matching end macro
appear before the encompassing block macro is either ended or subdivided.

The block macros are these:

@table @code
@item CASE
This macro has scope through the @code{ESAC} macro.
The scope is subdivided by @code{SELECT} macros.
You must have at least one @code{SELECT} macro.

@item DEFINE
This macro has scope through the @code{ENDDEF} macro.  The defined
user macro can never be a block macro.  This macro is extracted from
the template @i{before} the template is processed.  Consequently, you
cannot select a definition based on context.  You can, however, place
them all at the end of the file.

@item FOR
This macro has scope through the @code{ENDFOR} macro.

@item IF
This macro has scope through the @code{ENDIF} macro.
The scope may be subdivided by @code{ELIF} and @code{ELSE}
macros.  Obviously, there may be only one @code{ELSE} macro
and it must be the last of these subdivisions.

@item INCLUDE
This macro has the scope of the included file.
It is a block macro in the sense that the included
file must not contain any incomplete block macros.

@item WHILE
This macro has scope through the @code{ENDWHILE} macro.
@end table
@ignore
END   == MACROS == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == AUGMENTING == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c === SECTION MARKER

@node output controls
@section Redirecting Output
@cindex Redirecting Output
@cindex diversion

AutoGen provides a means for redirecting the template output to different
files or, in @file{M4} parlance, to various diversions.  It is accomplished
by providing a set of Scheme functions named @code{out-*}
(@pxref{AutoGen Functions}).

@table @samp
@item out-push-new (@pxref{SCM out-push-new})
This allows you to logically "push" output files onto a stack.
If you supply a string name, then a file by that name is created
to hold the output.  If you do not supply a name, then the text is
written to a scratch pad and retrieved by passing a ``@code{#t}'' argument
to the @code{out-pop} (@pxref{SCM out-pop}) function.

@item out-pop (@pxref{SCM out-pop})
This function closes the current output file and resumes output to the next
one in the stack.  At least one output must have been pushed onto the output
stack with the @code{out-push-new} (@pxref{SCM out-push-new}) function.  If
``@code{#t}'' is passed in as an argument, then the entire contents of the
diversion (or file) is returned.

@item out-suspend (@pxref{SCM out-suspend})
This function does not close the current output, but instead sets it aside
for resumption by the given name with @code{out-resume}.  The current output
must have been pushed on the output queue with @code{out-push-new}
(@pxref{SCM out-push-new}).

@item out-resume (@pxref{SCM out-resume})
This will put a named file descriptor back onto the top of
stack so that it becomes the current output again.

@item out-switch (@pxref{SCM out-switch})
This closes the current output and creates a new file,
purging any preexisting one.  This is a shortcut for "pop"
followed by "push", but this can also be done at the base level.

@item out-move (@pxref{SCM out-move})
Renames the current output file without closing it.
@end table

There are also several functions for determining the output
status.  @xref{AutoGen Functions}.

@ignore

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

@end ignore

@page
@node Augmenting AutoGen
@chapter Augmenting AutoGen Features
@cindex Augmenting AutoGen

AutoGen was designed to be simple to enhance.  You can do it by
providing shell commands, Guile/Scheme macros or callout functions
that can be invoked as a Guile macro.  Here is how you do these.

@menu
* shell commands::       Shell Output Commands
* guile macros::         Guile Macros
* guile callouts::       Guile Callout Functions
* AutoGen macros::       AutoGen Macros
@end menu

@c === SECTION MARKER

@node shell commands
@section Shell Output Commands

Shell commands are run inside of a server process.  This means that,
unlike @file{make}, context is kept from one command to the next.
Consequently, you can define a shell function in one place inside of
your template and invoke it in another.  You may also store values
in shell variables for later reference.  If you load functions from
a file containing shell functions, they will remain until AutoGen exits.

If your shell script should determine that AutoGen should stop
processing, the recommended method for stopping AutoGen is:
@example
die "some error text"
@end example

@noindent
That is a shell function added by AutoGen.  It will send a SIGTERM
to autogen and exit from the "persistent" shell.

@c === SECTION MARKER

@node guile macros
@section Guile Macros

Guile also maintains context from one command to the next.  This means you may
define functions and variables in one place and reference them elsewhere.
You also may load Guile macro definitions from a Scheme file by using the
@code{--load-scheme} command line option (@pxref{autogen load-scheme}).
Beware, however, that the AutoGen specific scheme functions have not been
loaded at this time, so though you may define functions that reference them,
do not invoke the AutoGen functions at this time.

If your Scheme script should determine that AutoGen should stop
processing, the recommended method for stopping AutoGen is:
@example
(error "some error text")
@end example

@c === SECTION MARKER

@node guile callouts
@section Guile Callout Functions

Callout functions must be registered with Guile to work.  This can
be accomplished either by putting your routines into a shared library
that contains a @code{void scm_init( void )} routine that registers
these routines, or by building them into AutoGen.

To build them into AutoGen, you must place your routines in the source
directory and name the files @file{exp*.c}.  You also must have a stylized
comment that @file{getdefs} can find that conforms to the following:

@example
/*=gfunc <function-name>
 *
 *  what:    <short one-liner>
 *  general_use:
 *  string:  <invocation-name-string>
 *  exparg:  <name>, <description> [, ['optional'] [, 'list']]
 *  doc:     A long description telling people how to use
 *           this function.
=*/
SCM
ag_scm_<function-name>( SCM arg_name[, ...] )
@{ <code> @}
@end example

@table @samp
@item gfunc
You must have this exactly thus.

@item <function-name>
This must follow C syntax for variable names

@item <short one-liner>
This should be about a half a line long.
It is used as a subsection title in this document.

@item general_use:
You must supply this unless you are an AutoGen maintainer and are writing
a function that queries or modifies the state of AutoGen.

@item <invocation-name-string>
Normally, the @code{function-name} string will be transformed into
a reasonable invocation name.  However, that is not always true.
If the result does not suit your needs, then supply an alternate string.

@item exparg:
You must supply one for each argument to your function.
All optional arguments must be last.
The last of the optional arguments may be a list, if you choose.

@item doc:
Please say something meaningful.

@item [, ...]
Do not actually specify an ANSI ellipsis here.  You must provide
for all the arguments you specified with @code{exparg}.
@end table

See the Guile documentation for more details.
More information is also available in a large comment at the
beginning of the @file{agen5/snarf.tpl} template file.

@c === SECTION MARKER

@node AutoGen macros
@section AutoGen Macros

There are two kinds@:  those you define yourself and AutoGen native.
The user-defined macros may be defined in your templates or loaded
with the @code{--lib-template} option
(See @ref{DEFINE} and  @ref{autogen lib-template}).

As for AutoGen native macros, do not add any. It is easy to do, but I
won't like it.  The basic functions needed to accomplish looping over
and selecting blocks of text have proved to be sufficient over a period
of several years.  New text transformations can be easily added via any
of the AutoGen extension methods, as discussed above.

@ignore
END   == AUGMENTING == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == INSTALLATION == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@page
@node Installation
@chapter Configuring and Installing

@menu
* configuring::    Configuring AutoGen
* AutoGen CGI::    AutoGen as a CGI server
* signal names::   Signal Names
* installing::     Installing AutoGen
@end menu

@c === SECTION MARKER

@node configuring
@section Configuring AutoGen
@cindex configuring

AutoGen is configured and built using Libtool, Automake and Autoconf.
Consequently, you can install it wherever you wish using the various
@samp{--prefix} options.  To the various configuration options supplied
by these tools, AutoGen adds a few of its own:

@table @samp
@item --disable-shell
AutoGen is now capable of acting as a CGI forms server, @xref{AutoGen CGI}.
As such, it will gather its definitions using either @samp{GET} or
@samp{POST} methods.  All you need to do is have a template named
@file{cgi.tpl} handy or specify a different one with a command line
option.

However, doing this without disabling the server shell brings
considerable risk.  If you were to pass user input to a script
that contained, say, the classic "@samp{`rm -rf /`}", you might have
a problem.  This configuration option will cause shell template
commands to simply return the command string as the result.
No mistakes.  Much safer.  Strongly recommended.
The default is to have server shell scripting enabled.

Disabling the shell will have some build side effects, too.

@itemize @bullet
@item
Many of the make check tests will fail, since they assume
a working server shell.
@item
The getdefs and columns programs are not built.
The options are distributed as definition files and they
cannot be expanded with a shell-disabled AutoGen.
@item
Similarly, the documentation cannot be regenerated because
the documentation templates depend on subshell functionality.
@end itemize

@item --enable-debug
Turning on AutoGen debugging enables very detailed inspection of
the input definitions and monitoring shell script processing.
These options are not particularly useful to anyone not directly
involved in maintaining AutoGen.  If you do choose to enable AutoGen
debugging, be aware that the usage page was generated without these
options, so when the build process reaches the documentation rebuild,
there will be a failure.  @samp{cd} into the @file{agen5} build
directory, @samp{make} the @samp{autogen.texi} file and all will
be well thereafter.

@item --with-regex-header
@itemx --with-header-path
@itemx --with-regex-lib
These three work together to specify how to compile with and link to
a particular POSIX regular expression library.  The value for
@file{--with-regex-header=value} must be the name of the relevant header
file.  The AutoGen sources will attempt to include that source with
a @code{#include <value>} C preprocessing statement.  The @code{path} from the
@file{--with-header-path=path} will be added to @code{CPPFLAGS} as @file{-Ipath}.
The @code{lib-specs} from @file{--with-regex-lib=lib-specs} will be added
to @code{LDFLAGS} without any adornment.
@end table

@c === SECTION MARKER

@page
@node AutoGen CGI
@section AutoGen as a CGI server

AutoGen is now capable of acting as a CGI forms server.
It behaves as a CGI server if the definitions input is from stdin
and the environment variable @code{REQUEST_METHOD} is defined
and set to either "GET" or "POST".  If set to anything else,
AutoGen will exit with a failure message.  When set to one of those
values, the CGI data will be converted to AutoGen definitions
(@pxref{Definitions File}) and the template named "@code{cgi.tpl}"
will be processed.

This works by including the name of the real template to process
in the form data and having the "@code{cgi.tpl}" template include
that template for processing.  I do this for processing the form
@url{http://autogen.sourceforge.net/conftest.html}.  The "@code{cgi.tpl}"
looks approximately like this:

@example
<? AutoGen5 Template ?>
<?
IF (not (exist? "template"))                       ?><?
  form-error                                       ?><?

ELIF (=* (get "template") "/")                     ?><?
  form-error                                       ?><?

ELIF (define tpl-file (string-append "cgi-tpl/"
                      (get "template")))
     (access? tpl-file R_OK)                       ?><?
  INCLUDE (. tpl-file)                             ?><?

ELIF (set! tpl-file (string-append tpl-file ".tpl"))
     (access? tpl-file R_OK)                       ?><?
  INCLUDE (. tpl-file)                             ?><?

ELSE                                               ?><?
  form-error                                       ?><?
ENDIF                                              ?>
@end example

@noindent
This forces the template to be found in the "@code{cgi-tpl/}"
directory.  Note also that there is no suffix specified in the
pseudo macro (@pxref{pseudo macro}).  That tells AutoGen to emit
the output to @file{stdout}.

The output is actually spooled until it is complete so that,
in the case of an error, the output can be discarded and a proper
error message can be written in its stead.

@strong{Please also note} that it is advisable, @emph{especially} for network
accessible machines, to configure AutoGen (@pxref{configuring}) with
shell processing disabled (@code{--disable-shell}).  That will make it
impossible for any referenced template to hand data to a subshell for
interpretation.

@c === SECTION MARKER

@node signal names
@section Signal Names
@cindex Signal Names

When AutoGen is first built, it tries to use @code{psignal(3)},
@code{sys_siglist}, @code{strsigno(3)} and @code{strsignal(3)} from the
host operating system.  If your system does not supply these, the
AutoGen distribution will.  However, it will use the distributed mapping
and this mapping is unlikely to match what your system uses.  This can
be fixed.  Once you have installed autogen, the mapping can be rebuilt
on the host operating system.  To do so, you must perform the
following steps:

@enumerate
@item
Build and install AutoGen in a place where it will be found in your
search path.

@item
@code{cd $@{top_srcdir@}/compat}

@item
@code{autogen strsignal.def}

@item
Verify the results by examining the @file{strsignal.h} file produced.

@item
Re-build and re-install AutoGen.
@end enumerate

If you have any problems or peculiarities that cause this process to
fail on your platform, please send me copies of the header files
containing the signal names and numbers, along with the full path names
of these files.  I will endeavor to fix it.  There is a shell script
inside of @file{strsignal.def} that tries to hunt down the information.

@c === SECTION MARKER

@node installing
@section Installing AutoGen
@cindex Installing

There are several files that get installed.  The number depend
whether or not both shared and archive libraries are to be
installed.  The following assumes that everything is installed
relative to @code{$prefix}.  You can, of course, use
@code{configure} to place these files where you wish.

@strong{NB}@:  AutoGen does not contain any compiled-in path names.
All support directories are located via option processing,
the environment variable @code{HOME} or finding the directory where
the executable came from.

The installed files are:

@enumerate
@item
The executables in @file{bin} (autogen, getdefs and columns).

@item
The AutoOpts link libraries as @file{lib/libopts.*}.

@item
An include file in @file{include/options.h}, needed for
Automated Option Processing (see next chapter).

@item
Several template files and a scheme script in @file{share/autogen}, needed
for Automated Option Processing (@pxref{AutoOpts}), parsing definitions
written with scheme syntax (@pxref{Dynamic Text}), the templates for
producing documentation for your program (@pxref{documentation attributes}),
autoconf test macros, and AutoFSM.

@item
Info-style help files as @file{info/autogen.info*}.
These files document AutoGen, the option processing
library AutoOpts, and several add-on components.

@item
The three man pages for the three executables are installed in man/man1.
@end enumerate

This program, library and supporting files can be installed
with three commands:

@itemize @bullet
@item
<src-dir>/configure [ <configure-options> ]
@item
make
@item
make install
@end itemize

However, you may wish to insert @code{make check}
before the @code{make install} command.

If you do perform a @code{make check} and there are any failures, you
will find the results in @code{<module>/test/FAILURES}.  Needless to say, I
would be interested in seeing the contents of those files and any
associated messages.  If you choose to go on and analyze one of these
failures, you will need to invoke the test scripts individually.  You
may do so by specifying the test (or list of test) in the TESTS make
variable, thus:

@example
gmake TESTS=test-name.test check
@end example

I specify @code{gmake} because most makes will not let you override
internal definitions with command line arguments.  @code{gmake} does.

All of the AutoGen tests are written to honor the contents of the
@t{VERBOSE} environment variable.  Normally, any commentary generated
during a test run is discarded unless the @t{VERBOSE} environment
variable is set.  So, to see what is happening during the test, you
might invoke the following with @i{bash} or @i{ksh}:

@example
VERBOSE=1 gmake TESTS="for.test forcomma.test" check
@end example

@noindent
Or equivalently with @i{csh}:

@example
env VERBOSE=1 gmake TESTS="for.test forcomma.test" check
@end example

@ignore
END   == INSTALLATION == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == AUTOFSM == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c === SECTION MARKER

@node AutoFSM
@section Automated Finite State Machine
@cindex AutoFSM
@cindex finite state machine

The templates to generate a finite state machine in C or C++ is included
with AutoGen.  The documentation is not.  The documentation is in HTML
format for @uref{http://www.gnu.org/software/autogen/autofsm.html,viewing},
or you can @uref{http://download.sourceforge.net/autogen/,download FSM}.

@node AutoXDR
@section Combined RPC Marshalling
@cindex RPC
@cindex rpcgen
@cindex remote procedure call
@cindex AutoXDR
@cindex XDR

The templates and NFSv4 definitions are not included with AutoGen in any way.
The folks that designed NFSv4 noticed that much time and bandwidth was
wasted sending queries and responses when many of them could be bundled.
The protocol bundles the data, but there is no support for it in rpcgen.
That means you have to write your own code to do that.  Until now.
Download this and you will have a large, complex example of how to use
@code{AutoXDR} for generating the marshaling and unmarshaling of combined
RPC calls.  There is a brief example
@uref{http://www.gnu.org/software/autogen/xdr/index.html,on the web}, but
you should @uref{http://download.sourceforge.net/autogen/,download AutoXDR}.

@c === SECTION MARKER

@node AutoEvents
@section Automated Event Management
@cindex AutoEvents

Large software development projects invariably have a need to manage
the distribution and display of state information and state changes.
In other words, they need to manage their software events.  Generally,
each such project invents its own way of accomplishing this and then
struggles to get all of its components to play the same way.  It is a
difficult process and not always completely successful.  This project
helps with that.

AutoEvents completely separates the tasks of supplying the data
needed for a particular event from the methods used to manage the
distribution and display of that event.  Consequently, the programmer
writing the code no longer has to worry about that part of the
problem.  Likewise the persons responsible for designing the event
management and distribution no longer have to worry about getting
programmers to write conforming code.

This is a work in progress.  See my
@uref{http://www.gnu.org/software/autogen/autoevents.html,web page}
on the subject, if you are interested.
I have some useful things put together, but it is not ready
to call a product.

@ignore
END   == AUTOFSM == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
@c
@c  AUTOOPTS SECTIONS
@c
@c
@ignore
START == AUTOOPTS == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

All the features notwithstanding, some applications simply have
well-established command line interfaces.  Even still, those programs
may use the configuration file parsing portion of the library.
See the ``AutoOpts Features'' and ``Configuration File Format'' sections.

@menu
* Features::            AutoOpts Features
* Licensing::           AutoOpts Licensing
* Caveats::             Developer and User Notes
* Quick Start::         Quick Start
* Option Definitions::  Option Definitions
* AutoOpts API::        Programmatic Interface
* Multi-Threading::     Multi-Threading
* option descriptor::   Option Descriptor File
* Using AutoOpts::      Using AutoOpts
* Presetting Options::  Configuring your program
* Config File Format::  Configuration File Format
* shell options::       AutoOpts for Shell Scripts
* AutoInfo::            Automated Info Docs
* AutoMan pages::       Automated Man Pages
* getopt_long::         Using getopt(3C)
* i18n::                Internationalizing AutoOpts
* Naming Conflicts::    Naming Conflicts
* All Attribute Names:: All Attribute Names
* Option Define Names:: Option Definition Name Index
@end menu

@c === SECTION MARKER

@node Features
@section AutoOpts Features
@cindex features

AutoOpts supports option processing; option state saving; and
program documentation with innumerable features.  Here, we list
a few obvious ones and some important ones, but the full list is
really defined by all the attributes defined in the @ref{Option Definitions}
section.

@enumerate
@item
POSIX-compliant short (flag) option processing.

@item
GNU-style long options processing.  Long options
are recognized without case sensitivity, and they may be abbreviated.

@item
Environment variable initializations, @xref{environrc}.

@item
Initialization from configuration files (aka RC or INI files), and
saving the option state back into one, @xref{loading rcfile}.

@item
Config files may be partitioned.  One config file may be used by several
programs by partitioning it with lines containing,
``@code{[PROGRAM_NAME]}'' or ``@code{<?program-name>}'', @xref{loading rcfile}.

@item
Config files may contain AutoOpts directives.
``@code{<?auto-options [[option-text]]>}'' may be used to set @code{AutoOpts}
option processing options.  Viz., @code{GNU} usage layout versus @code{AutoOpts}
conventional layout, and @code{misuse-usage} versus @code{no-misuse-usage},
@xref{usage attributes}.

@item
Options may be marked as @code{@i{dis}-abled} with a disablement prefix.
Such options may default to either an enabled or a disabled state.  You
may also provide an enablement prefix, too, e.g., @code{--allow-mumble}
and @code{--prevent-mumble} (@pxref{Common Attributes}).

@item
Verify that required options are present between the minimum and maximum
number of times on the command line.  Verify that conflicting options do not
appear together.  Verify that options requiring the presence of other options
are, in fact, used in the presence of other options.
See @xref{Common Attributes}, and @xref{Option Conflict Attributes}.

@item
There are several @ref{automatic options, automatically supported options}.
They will have short flags if any options have option flags and the flags
are not suppressed.  The associated flag may be altered or suppressed by
specifying no value or an alternate character for ``@code{xxx-value;}'' in
the option definition file.  ``@code{xxx}'' is the name of the option below:

@table @samp
@item --help
@itemx --more-help
These are always available.  @samp{--more-help} will pass the full usage
text through a pager.
@item --usage
@vindex usage-opt
This is added to the option list if @code{usage-opt} is specified.
It yields the abbreviated usage to @file{stdout}.
@item --version
This is added to the option list if @code{version = xxx;} is specified.
@item --load-opts
@itemx --save-opts
These are added to the option list if @code{homerc} is specified.  Mostly.
If, @code{disable-save} is specified, then @samp{--save-opts} is disabled.
@end table

@item
Various forms of main procedures can be added to the output,
@xref{Generated main}.  There are four basic forms:

@enumerate a
@item
A program that processes the arguments and writes to standard out
portable shell commands containing the digested options.

@item
A program that will generate portable shell commands to parse the defined
options.  The expectation is that this result will be copied into a
shell script and used there.

@item
A ``for-each'' main that will invoke a named function once for either
each non-option argument on the command line or, if there are none,
then once for each non-blank, non-comment input line read from stdin.

@item
A main procedure of your own design.  Its code can be supplied in the
option description template or by incorporating another template.
@end enumerate

@item
There are several methods for handling option arguments.
@itemize @bullet
@item
nothing (@pxref{OPT_ARG}) option argument strings are globally available.
@item
user supplied (@pxref{Option Argument Handling})
@item
stack option arguments (@pxref{Option Argument Handling})
@item
integer numbers (@pxref{arg-type number})
@item
true or false valued (@pxref{arg-type boolean})
@item
enumerated list of names (@pxref{arg-type keyword})
@item
an enumeration (membership) set (@pxref{arg-type set membership})
@item
a list of name/value pairs (option ``subopts'') (@pxref{arg-type hierarchy})
@item
a time duration or a specific time and date
@item
validated file name (@pxref{arg-type file name})
@item
optional option argument (@pxref{arg-optional})
@end itemize

@item
The generated usage text can be emitted in either AutoOpts standard
format (maximizing the information about each option), or GNU-ish
normal form.  The default form is selected by either specifying or not
specifying the @code{gnu-usage} attribute (@pxref{information attributes}).
This can be overridden by the user himself with the
@code{AUTOOPTS_USAGE} environment variable.  If it exists and is set
to the string @code{gnu}, it will force GNU-ish style format; if it is
set to the string @code{autoopts}, it will force AutoOpts standard
format; otherwise, it will have no effect.

@item
The usage text and many other strings are stored in a single character array
(@pxref{SCM string-table-new,string table functions}).  This reduces fixup
costs when loading the program or library.  The downside is that if GCC
detects that any of these strings are used in a printf format, you may get
the warning, @code{embedded '\0' in format}.  To eliminate the warning, you
must provide GCC with the @code{-Wno-format-contains-nul} option.

@item
If you compile with @code{ENABLE_NLS} defined and @code{_()} defined to
a localization function (e.g. @code{gettext(3GNU)}), then the option
processing code will be localizable (@pxref{i18n}).  Provided also that
you do not define the @code{no-xlate} attribute to @emph{anything}
(@pxref{presentation attributes}).

@item
Provides a callable routine to parse
a text string as if it were from one of the rc/ini/config files,
hereafter referred to as a configuration file.

@item
By adding a @samp{doc} and @samp{arg-name} attributes to each option,
AutoGen will also be able to produce a man page and the @samp{invoking}
section of a texinfo document.

@item
Intermingled option processing.  AutoOpts options may be intermingled with
command line operands and options processed with other parsing techniques.
This is accomplished by setting the @code{allow-errors}
(@pxref{program attributes}) attribute.  When processing reaches a point
where @code{optionProcess} (@pxref{libopts-optionProcess}) needs to be called
again, the current option can be set with @code{RESTART_OPT(n)}
(@pxref{RESTART_OPT}) before calling @code{optionProcess}.

See: @xref{library attributes}.

@item
Library suppliers can specify command line options that their
client programs will accept.  They specify option definitions
that get @code{#include}-d into the client option definitions
and they specify an "anchor" option that has a callback and must be invoked.
That will give the library access to the option state for their options.

@item
library options.  An AutoOpt-ed library may export its options for use in
an AutoOpt-ed program.  This is done by providing an option definition file
that client programs @code{#include} into their own option definitions.
See ``AutoOpt-ed Library for AutoOpt-ed Program'' (@pxref{lib and program})
for more details.
@end enumerate

@c === SECTION MARKER

@node Licensing
@section AutoOpts Licensing
@cindex Licensing

When AutoGen is installed, the AutoOpts project is installed with it.
AutoOpts includes various AutoGen templates and a pair of shared
libraries.  These libraries may be used under the terms of version 3
of the GNU Lesser General Public License (LGPL).

One of these libraries (@code{libopts}) is needed by programs that are built
using AutoOpts generated code.  This library is available as a separate
``tear-off'' source tarball.  It is redistributable for use under either of
two licenses: The above mentioned GNU Lesser General Public License, and
the advertising-clause-free BSD license.  Both of these license terms are
incorporated into appropriate COPYING files included with the @code{libopts}
source tarball.  This source may be incorporated into your package with
the following simple commands:

@example
rm -rf libopts libopts-*
gunzip -c `autoopts-config libsrc` | \
   tar -xvf -
mv libopts-*.*.* libopts
@end example

View the @file{libopts/README} file for further integration information.

@c === SECTION MARKER

@page
@node Caveats
@section Developer and User Notes

AutoOpts has its conventional way of displaying option information
that includes somewhat more information that the standard GNU method.
AutoOpts will also print out a line of usage text for each option type
when options are misspecified.  GNU programs typically do not do this.
These defaults can be changed on a per-program basis by adding either
or both of the following in the option definition file:

@example
gnu-usage;
no-misuse-usage;
@end example

Users may also override these settings with the @code{AUTOOPTS_USAGE}
environment variable.  It may be set to a comma or white space separated
list of the following strings:

@table @samp
@item gnu
@cindex gnu
The format of the extended usage text will be displayed in GNU-normal form.
The default display for @code{--version} will be to include a note
on licensing terms.

@item autoopts
@cindex autoopts
The format of the extended usage will be in AutoOpts' native layout.

@item no-misuse-usage
@cindex no-misuse-usage
When an option error is made on the command line, the abbreviated
usage text will be suppressed.

@item misuse-usage
@cindex misuse-usage
When an option error is made on the command line, the abbreviated
usage text will be shown.
@end table

@noindent
The setting used is the last one seen.  The @code{autoopts} and
@code{misuse-usage} serve no purpose, unless the definition file
entries were specified as above.

@b{Note for developers}:

The templates used to implement AutoOpts depend heavily upon
token pasting.  That mens that if you name an option, ``debug'', for
example, the generated header will expect to be able to emit
@code{#define} macros such as this:
@example
#define DESC(n) (autogenOptions.pOptDesc[INDEX_OPT_## n])
@end example
and expect @code{DESC(DEBUG)} to expand correctly into
@code{(autogenOptions.pOptDesc[INDEX_OPT_DEBUG])}.
If @code{DEBUG} is @code{#defined} to something else, then
that something else will be in the above expansion.

If you discover you are having strange problems like this,
you may wish to use some variation of the @code{guard-option-names}
@xref{program attributes}.


@c === SECTION MARKER

@page
@node Quick Start
@section Quick Start
@cindex example, simple AutoOpts

Since it is generally easier to start with a simple example than it is
to look at the options that AutoGen uses itself, here is a very simple
AutoOpts example.  You can copy this example out of the Info file and
into a source file to try it.  You can then embellish it into what you
really need.  For more extensive examples, you can also examine the help
output and option definitions for the commands @code{columns},
@code{getdefs} and @code{autogen} itself.

For our simple example, assume you have a program named @code{check}
that takes two options:

@enumerate
@item
A list of directories to check over for whatever it is @code{check} does.
You want this option available as a POSIX-style flag option
and a GNU long option.  You want to allow as many of these
as the user wishes.
@item
An option to show or not show the definition tree being used.
Only one occurrence is to be allowed, specifying one or the other.
@end enumerate

@ignore
END   == AUTOOPTS == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == AUTOOPTS-MAIN == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@noindent
Normally, however, you would not use the ``main'' clause.  Instead,
the file would be named something like @file{checkopt.def}, you would
compile @file{checkopt.c} the usual way, and link the object with the rest
of your program.

The options are processed by calling @code{optionProcess}
(@pxref{libopts-optionProcess}):

@example
main( int argc, char** argv )
@{
  @{
    int optct = optionProcess( &checkOptions, argc, argv );
    argc -= optct;
    argv += optct;
  @}
@end example

The options are tested and used as in the following fragment.
``@code{ENABLED_OPT}'' is used instead of ``@code{HAVE_OPT}'' for the
@code{show-defs} option because it is an enabled/disabled option type:

@example
  if (  ENABLED_OPT( SHOW_DEFS )
     && HAVE_OPT( CHECK_DIRS )) @{
    int    dirct = STACKCT_OPT( CHECK_DIRS );
    char** dirs  = STACKLST_OPT( CHECK_DIRS );
    while (dirct-- > 0) @{
      char* dir = *dirs++;
      ...
@end example

The ``doc'' clauses are used in the flag stanzas for man pages and
texinfo invoking documentation.  With the above definition file, the
two following commands will produce the two documentation files
@file{check.1} and @file{invoke-check.texi}.  The latter file will
be generated as a chapter, rather than a section or subsection.

@example
autogen -Tagman-cmd check.def
autogen -DLEVEL=chapter -Tagtexi-cmd -binvoke-check.texi check.def
@end example

@noindent
The result of which is left as an exercise for the reader.

A lot of magic happens to make this happen.
The rest of this chapter will describe the myriad of option attributes
supported by AutoOpts.  However, keep in mind that, in general, you won't
need much more than what was described in this "quick start" section.

@node Option Definitions
@section Option Definitions
@cindex Option Definitions

AutoOpts uses an AutoGen definitions file for the definitions of the
program options and overall configuration attributes.
The complete list of program and option attributes is quite extensive,
so if you are reading to understand how to use AutoOpts, I recommend
reading the "Quick Start" section (@pxref{Quick Start}) and paying
attention to the following:

@enumerate
@item
@code{prog-name}, @code{prog-title}, and @code{argument}, program
attributes, @xref{program attributes}.
@item
@code{name} and @code{descrip} option attributes, @xref{Required Attributes}.
@item
@code{value} (flag character) and @code{min} (occurrence counts)
option attributes, @xref{Common Attributes}.
@item
@code{arg-type} from the option argument specification section,
@xref{Option Arguments}.
@item
Read the overall how to, @xref{Using AutoOpts}.
@item
Highly recommended, but not required, are the several "man" and
"info" documentation attributes, @xref{documentation attributes}.
@end enumerate

Keep in mind that the majority are rarely used and can be safely
ignored.  However, when you have special option processing requirements,
the flexibility is there.

@menu
* program attributes::          Program Description Attributes
* library attributes::          Options for Library Code
* information attributes::      Program Information Attributes
* Generated main::              Generating main procedures
* option attributes::           Option Attributes
* Option Arguments::            Option Argument Specification
* Option Argument Handling::    Option Argument Handling
* Internationalizing Options::  Internationalizing Options
* documentation attributes::    Man and Info doc Attributes
* automatic options::           Automatically Supported Options
* standard options::            Library of Standard Options
@end menu

@node program attributes
@subsection Program Description Attributes
@cindex program attributes

The following global definitions are used to define attributes of the entire
program.  These generally alter the configuration or global behavior of the
AutoOpts option parser.  The first two are required of every program.  The
third is required if there are to be any left over arguments (operands)
after option processing.  The rest have been grouped below.  Except as noted,
there may be only one copy of each of these definitions:

@table @samp

@item prog-name
@vindex prog-name
This attribute is required.  Variable names derived from this name
are derived using @code{string->c_name!} (@pxref{SCM string->c-name!}).

@item prog-title
@vindex prog-title
This attribute is required and may be any descriptive text.

@item argument
@vindex argument
This attribute is required if your program uses operand arguments.
It specifies the syntax of the arguments that @strong{follow} the options.
It may not be empty, but if it is not supplied, then option processing
must consume all the arguments.  If it is supplied and starts with an
open bracket (@code{[}), then there is no requirement on the presence or
absence of command line arguments following the options.  Lastly, if it
is supplied and does not start with an open bracket, then option
processing must @strong{not} consume all of the command line arguments.

@item config-header
@vindex config-header
If your build has a configuration header, it must be included before
anything else.  Specifying the configuration header file name with this
attribute will cause that to happen.
@end table

@menu
* usage attributes::            Usage and Version Info Display
* config attributes::           Program Configuration
* programming attributes::      Programming Details
* presentation attributes::     User Presentation Attributes
@end menu

@c
@c = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
@c
@node usage attributes
@subsubsection Usage and Version Info Display

These will affect the way usage is seen and whether or not version
information gets displayed.

@table @samp
@item full-usage
@vindex full-usage
If this attribute is provided, it may specify the full length
usage text, or a variable name assignable to a ``char const *'' pointer,
or it may be empty.  The meanings are determined by  the length.
@itemize @bullet
@item
If not provided, the text will be computed as normal.
@item
If the length is zero, then the usage text will be derived from
the current settings and inserted as text into the generated .c file.
@item
If the length is 1 to 32 bytes, then it is presumed to be a variable
name that either points to or is an array of const chars.
@item
If it is longer than that, it is presumed to be the help text itself.
This text will be inserted into the generated .c file.
@end itemize

This string should be readily translatable.  Provision will be made
to translate it if this is provided, if the source code is compiled with
@code{ENABLE_NLS} defined, and @code{no-xlate} has not been set to the
value @emph{anything}.

@item short-usage
@vindex short-usage
If this attribute is provided, it is used to specify an abbreviated
version of the usage text.  This text is constructed in the same way
as the ``full-usage'', described above.

@item gnu-usage
@vindex gnu-usage
AutoOpts normaly displays usage text in a format that provides more
information than the standard GNU layout, but that also means it is
not the standard GNU layout.  This attribute changes the default to
GNU layout, with the @code{AUTOOPTS_USAGE} environment variable used
to request @code{autoopts} layout.
See @xref{Caveats, Developer and User Notes}.

@item usage-opt
@vindex usage-opt
I apologize for too many confusing usages of usage.
This attribute specifies that @code{--usage} and/or @code{-u} be
supported.  The help (usage) text displayed will be abbreviated
when compared to the default help text.

@item no-misuse-usage
@vindex no-misuse-usage
When there is a command line syntax error, by default AutoOpts will
display the abbreviated usage text, rather than just a one line
``you goofed it, ask for usage'' message.  You can change the default
behavior for your program by supplying this attribute.  The user may
override this choice, again, with the @code{AUTOOPTS_USAGE} environment
variable.  See @xref{Caveats, Developer and User Notes}.

@item prog-group
@vindex prog-group
The version text in the @file{getopt.tpl} template will include this
text in parentheses after the program name, when this attribute is specified.
For example:
@example
mumble (stumble) 1.0
@end example
@noindent
says that the ``@code{mumble}'' program is version 1.0 and is part of the
``@code{stumble}'' group of programs.

@item usage
@vindex usage
If your program has some cleanup work that must be done before exiting
on usage mode issues, or if you have to customize the usage message in
some way, specify this procedure and it will be called instead of the
default @code{optionUsage()} function.  For example, if a program is
using the curses library and needs to invoke the usage display, then
you must arrange to call @code{endwin()} before invoking the library
function @code{optionUsage()}.  This can be handled by specifying your
own usage function, thus:
@example
void
my_usage(tOptions * opts, int ex)
@{
    if (curses_window_active)
        endwin();
    optionUsage(opts, ex);
@}
@end example

@item version
@vindex version
Specifies the program version and activates the VERSION option,
@xref{automatic options}.
@end table

@c
@c = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
@c
@node config attributes
@subsubsection Program Configuration

Programs may be ``pre-configured'' before normal command line options
are processed (See @pxref{Immediate Action, Immediate Action Attributes}).
How configuration files and environment variables are handled get
specified with these attributes.

@table @samp
@item disable-load
@itemx disable-save
@vindex disable-load
@vindex disable-save
Indicates that the command line usage of @code{--load-opts} and/or
@code{--save-opts} are disallowed.

@item environrc
@vindex environrc
Indicates looking in the environment for values of variables named,
@code{PROGRAM_OPTNAME} or @code{PROGRAM}, where @code{PROGRAM} is the
upper cased @code{C-name} of the program and @code{OPTNAME} is the
upper cased @code{C-name} of a specific option.  The contents of
the @code{PROGRAM} variable, if found, are tokenized and processed.
The contents of @code{PROGRAM_OPTNAME} environment variables are taken
as the option argument to the option nameed @code{optname}.

@item homerc
@vindex homerc
Specifies that option settings may be loaded from and stored into
configuration files.  Each instance of this attribute is either a directory or
a file using a specific path, a path based on an environment variable or a
path relative to installation directories.  The method used depends on the name.
If the one entry is empty, it enables the loading and storing of settings,
but no specific files are searched for.  Otherwise, a series of configuration
files are hunted down and, if found, loaded.

If the first character of the @samp{homerc} value is not the dollar
character (@code{$}), then it is presumed to be a path name based on the
current directory.  Otherwise, the method depends on the second character:

@table @code
@item $
The path is relative to the directory where the executable was found.
@item @@
The path is relative to the package data directory, e.g.
@code{/usr/local/share/autogen}.
@item [a-zA-Z]
The path is derived from the named environment variable.
@end table

Use as many as you like.  The presence of this attribute
activates the @code{--save-opts} and @code{--load-opts} options.
However, saving into a file may be disabled with the @samp{disable-save}.
@xref{loading rcfile}.
See the @code{optionMakePath(3AGEN)} man page for excruciating details.

@item rcfile
@vindex rcfile
Specifies the configuration file name.  This is only useful if you
have provided at least one @code{homerc} attribute.
@example
default: .<prog-name>rc
@end example

@item vendor-opt
@vindex vendor-opt
This option implements the @code{-W} vendor option command line option.

For POSIX specified utilities, the options are constrained to the options
that are specified by POSIX.  Extensions should be handled with @code{-W}
command line options, the short flag form.  Long option name processing
must be disabled.  In fact, the @code{long-opts} attribute must not be
provided, and some options must be specified without flag values.

The @code{-W long-name} is processed by looking up the long option
name that follows it.  It cannot be a short flag because that would
conflict with the POSIX flag name space.  It will be processed as if
long options were accepted and @code{--long-name} were found on the
command line.
@end table

@c
@c = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
@c
@node programming attributes
@subsubsection Programming Details

These attributes affect some of the ways that the option data are
used and made available to the program.

@table @samp
@item config-header
@vindex config-header
The contents of this attribute should be just the name of the configuration
file.  A "#include" naming this file will be inserted at the top of the
generated header.

@item exit-name
@itemx exit-desc
@vindex exit-name
@vindex exit-desc
These values should be defined as indexed values, thus:
@example
exit-name[0] = success;
exit-desc[0] = 'Successful program execution.';
exit-name[1] = failure;
exit-desc[1] = 'The operation failed or command syntax was not valid.';
@end example
By default, all programs have these effectively defined for them.
They may be overridden by explicitly defining any or all of these values.
Additional names and descriptions may be defined.
They will cause an enumeration to be emitted, like this one
for @code{getdefs}:
@example
typedef enum @{
    GETDEFS_EXIT_SUCCESS = 0,
    GETDEFS_EXIT_FAILURE = 1
@} getdefs_exit_code_t;
@end example
@noindent
which will be augmented by any @code{exit-name} definitions beyond ``1''.

@item usage-message
@vindex usage-message
This attribute will cause two procedures to be added to the code file:
@code{usage_message} and @code{vusage_message}, with any applicable prefix
(see @code{prefix}, below).  They are declared in the
generated header, thus:
@example
extern void vusage_message(char const * fmt, va_list ap);
extern void usage_message(char const * fmt, ...);
@end example
@noindent
These functions print the message to @file{stderr} and invoke the usage
function with the exit code set to @code{1} (@code{EXIT_FAILURE}).

@item die-code
@vindex die-code
This tells AutoOpts templates to emit code for @code{vdie}, @code{die} and
@code{fserr} functions.  If the @code{die-code} is assigned a text value,
then that code will be inserted in the @code{vdie} function immediately
before it prints the death rattle message.

The profiles for these functions are:
@example
extern void vdie( int exit_code, char const * fmt, va_list);
extern void die(  int exit_code, char const * fmt, ...);
extern void fserr(int exit_code, char const * op, char const * fname);
@end example

@item export
@vindex export
This string is inserted into the .h interface file.  Generally used for
global variables or @code{#include} directives required by
@code{flag-code} text and shared with other program text.
Do not specify your configuration header (@file{config.h}) in this
attribute or the @code{include} attribute, however.  Instead, use
@code{config-header}, above.

@item guard-option-names
@vindex guard-option-names
AutoOpts generates macros that presume that there are no @code{cpp} macros
with the same name as the option name.  For example, if you have an option
named, @code{debug}, then you must not use @code{#ifdef DEBUG} in your code.
If you specify this attribute, every option name will be guarded.  If the name
is @code{#define}-d, then a warning will be issued and the name undefined.
If you do not specify this and there is a conflict, you will get strange
error messages.

This attribute may be set to any of four recognized states:

@itemize @bullet
@item
Not defined.  AutoOpts will behave as described above.

@item
Defined, but set to the empty string.  Text will be emitted into the header
to undefine (@code{#undef}) any conflicting preprocessor macros.  The code
will include compiler warnings (via @code{#warning}).  Some compilers are
not ANSI-C-99 compliant yet and will error out on those warnings.  You may
compile with @code{-DNO_OPTION_NAME_WARNINGS} to silence or mostly silence
them.

@item
Defined and set to the string, ``@code{no-warning}''.  All of the needed
@code{#undef}s will be emitted, without any conflict checking @code{#warning}
directives emitted.

@item
Defined and set to the string, ``@code{full-enum}''.  The option manipulation
preprocessor macros will not token paste the option names to the index
enumeration prefix.  e.g. you will need to use @code{HAVE_OPT(INDEX_OPT_DEBUG)}
instead of @code{HAVE_OPT(DEBUG)}.
@end itemize

@item include
@vindex include
This string is inserted into the .c file.  Generally used for global
variables required only by @code{flag-code} program text.

@item no-libopts
@vindex no-libopts
If you are going to handle your option processing with the @code{getopt.tpl}
template instead of using libopts, then specify this attribute.  It will
suppress mention of @code{--more-help} in the generated documentation.
(@code{getopt_long} does not support @code{--more-help}.)

@item prefix
@vindex prefix
This value is inserted into @strong{all} global names.  This will
disambiguate them if more than one set of options are to be compiled
into a single program.
@end table

@c
@c = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
@c
@node presentation attributes
@subsubsection User Presentation Attributes

Attributes that affect the user's experience.

@table @samp
@item allow-errors
@vindex allow-errors
The presence of this attribute indicates ignoring any command line
option errors.  This may also be turned on and off by invoking the
macros @code{ERRSKIP_OPTERR} and @code{ERRSTOP_OPTERR} from the
generated interface file.

@item long-opts
@vindex long-opts
@cindex named option mode
Presence indicates GNU-standard long option processing.  Partial name
matches are accepted, if they are at least two characters long and the
partial match is unique.  The matching is not case sensitive, and the
underscore, hyphen and carat characters are all equivalent (they match).

If any options do not have an option value (flag character) specified,
and least one does specify such a value, then you must specify
@code{long-opts}.  If none of your options specify an option value
(flag character) and you do not specify @code{long-opts}, then command
line arguments are processed in "named option mode".  This means that:

@itemize @bullet
@item
Every command line argument must be a long option.
@item
The flag markers @code{-} and @code{--} are completely optional.
@item
The @code{argument} program attribute is disallowed.
@item
One of the options may be specified as the default
(as long as it has a required option argument).
@end itemize

@item no-xlate
@vindex no-xlate
Modifies when or whether option names get translated.  If provided,
it must be assigned one of these values:
@table @samp
@item opt-cfg
to suppress option name translation for configuration file and and environment
variable processing.
@item opt
to suppress option name translation completely.  The usage text will
always be translated if @code{ENABLE_NLS} is defined and you have
translations for that text.
@item anything
Specifies disabling all internationalization support for option code, completely.
@end table
See also the various @code{XLAT} interface entries in the
AutoOpts Programmatic Interface section (@pxref{AutoOpts API}).

@item reorder-args
@vindex reorder-args
Normally, POSIX compliant commands do not allow for options to be interleaved
with operands.  If this is necessary for historical reasons, there are two
approaches available:
@itemize @bullet
@item
Allow @code{optionProcess} to return the index of the operand like it normally
does and process the operand(s).  When an operand is encountered that starts
with a hyphen, then set the AutoOpts current index with the @code{RESTART_OPT}
macro (see @pxref{RESTART_OPT}), and re-invoke @code{optionProcess}.  This will
also allow you to process the operands in context.

@item
Specify this attribute.  AutoOpts will re-order the command arguments
so that the operands appear (in the original order) at the end of
the argument list.  Differing configuration state is not possible
to detect after all options have been processed.
@end itemize

@item resettable
@vindex resettable
Specifies that the @code{--reset-option} command line option is to be supported.
This makes it possible to suppress any setting that might be found in
a configuration file or environment variable.
@end table

@node library attributes
@subsection Options for Library Code
@cindex library attributes

Some libraries provide their own code for processing command line
options, and this may be used by programs that utilize AutoOpts.
You may also wish to write a library that gets configured with AutoOpts
options and config files.  Such a library may either supply its own
configury routine and process its own options, or it may export its
option descriptions to programs that also use AutoOpts.  This section
will describe how to do all of these different things.

@menu
* lib and program::         AutoOpt-ed Library for AutoOpt-ed Program
* lib called::              AutoOpt-ed Library for Regular Program
* prog calls lib::          AutoOpt-ed Program Calls Regular Library
@end menu

@node lib and program
@subsubsection AutoOpt-ed Library for AutoOpt-ed Program

The library source code must provide an option definition file that consists
of only the attribute @code{library}
@vindex library
and @code{flag} entries.  The ``library'' attribute does not need any
associated value, so it will generally appeary by itself on a line folowed
by a semi-colon.  The first @code{flag} entry must contain the following
attributes:

@table @samp
@item name
This name is used in the construction of a global pointer of type
@code{tOptDesc const*}.  It is always required.
@item documentation
@vindex documentation
It tells @code{AutoOpts} that this option serves no normal purpose.
It will be used to add usage clarity and to locate option descriptors
in the library code.
@item descrip
This is a string that is inserted in the extended usage display
before the options specific to the current library.  It is always required.
@item lib-name
@vindex lib-name
This should match the name of the library.  This string is also used in
the construction of the option descriptor pointer name.  In the end, it
looks like this:
@example
extern tOptDesc const* <<lib-name>>_<<name>>_optDesc_p;
@end example
@noindent
and is used in the macros generated for the library's @code{.h} file.
@end table

In order to compile this @code{AutoOpts} using library, you must create a
special header that is not used by the client program.  This is accomplished
by creating an option definition file that contains essentially exactly the
following:

@example
AutoGen definitions options;
prog-name  = does-not-matter;  // but is always required
prog-title = 'also does not matter';  // also required
config-header = 'config.h'; // optional, but common
library;
#include library-options-only.def
@end example

@noindent
and nothing else.  AutoGen will produce only the @code{.h} file.
You may now compile your library, referencing just this @code{.h} file.
The macros it creates will utilize a global variable that will be defined
by the @code{AutoOpts}-using client program.  That program will need to
have the following @code{#include} in @i{its} option definition file:

@example
#include library-options-only.def
@end example

@noindent
All the right things will magically happen so that the global variables
named @code{<<lib-name>>_<<name>>_optDesc_p} are initialized correctly.
For an example, please see the @code{AutoOpts} test script:
@file{autoopts/test/library.test}.

@node lib called
@subsubsection AutoOpt-ed Library for Regular Program

In this case, your library must provide an option processing function
to a calling program.  This is accomplished by setting the @code{allow-errors}
global option attribute.  Each time your option handling function is called,
you must determine where your scan is to resume and tell the AutoOpts library
by invoking:

@example
RESTART_OPT(next_arg_index);
@end example

@noindent
and then invoke @code{not_opt_index = optionProcess(...)}.
The @code{not_opt_index} value can be used to set @code{optind},
if that is the global being used to scan the program argument array.

In this method, do @strong{NOT} utilize the global @code{library} attribute.
Your library must specify its options as if it were a complete program.
You may choose to specify an alternate @code{usage()} function so that
usage for other parts of the option interface may be displayed as well.
See ``Program Information Attributes'' (@pxref{information attributes}).

At the moment, there is no method for calling @code{optionUsage()} telling
it to produce just the information about the options and not the program
as a whole.  Some later revision after somebody asks.

@node prog calls lib
@subsubsection AutoOpt-ed Program Calls Regular Library

As with providing an @code{AutoOpt}-ed library to a non-@code{AutoOpt}-ed
program, you must write the option description file as if you were writing
all the options for the program, but you should specify the
@code{allow-errors} global option attribute and you will likely want an
alternate @code{usage()} function (see ``Program Information Attributes''
@pxref{information attributes}).  In this case, though, when
@code{optionProcess()} returns, you need to test to see if there might be
library options.  If there might be, then call the library's exported
routine for handling command line options, set the next-option-to-process
with the @code{RESTART_OPT()} macro, and recall @code{optionProcess()}.
Repeat until done.

@node information attributes
@subsection Program Information Attributes
@cindex information attributes

These attributes are used to define how and what information is displayed
to the user of the program.

@table @samp
@item copyright
@vindex copyright
The @code{copyright} is a structured value containing three to five
values.  If @code{copyright} is used, then the first three are required.

@enumerate
@item
@vindex date
@file{date} - the list of applicable dates for the copyright.
@item
@vindex owner
@file{owner} - the name of the copyright holder.
@item
@vindex type
@file{type} - specifies the type of distribution license.
AutoOpts/AutoGen supports the text of the GNU Public License (@file{gpl}),
the GNU ``Lesser'' General Public License with Library extensions
(@file{lgpl}), the Modified Free BSD license (@file{mbsd}) and a few others.
Other licenses may be specified, but you must provide your own license file.
The list of license files provided by AutoOpts may be seen by typing:
@example
ls $(autoopts-config pkgdatadir)/*.lic
@end example
@item
@vindex text
@file{text} - the text of the copyright notice.  This must be provided
if @file{type} is set to @file{NOTE}.
@item
@vindex author
@file{author} - in case the author name is to appear in the documentation
and is different from the copyright owner.
@item
@vindex eaddr
@file{eaddr} - email address for receiving praises and complaints.
Typically that of the author or copyright holder.
@end enumerate
@*
An example of this might be:
@example
copyright = @{
    date  = "1992-2012";
    owner = "Bruce Korb";
    eaddr = 'bkorb@@gnu.org';
    type  = GPL;
@};
@end example

@item detail
@vindex detail
This string is added to the usage output when the HELP option is
selected.

@item explain
@vindex explain
Gives additional information whenever the usage routine is invoked.

@item package
@vindex package
The name of the package the program belongs to.  This will appear
parenthetically after the program name in the version and usage output,
e.g.:  @code{autogen @i{(GNU autogen)} - The Automated Program Generator}.

@item preserve-case
@vindex preserve-case
This attribute will not change anything except appearance.  Normally, the
option names are all documented in lower case.  However, if you specify this
attribute, then they will display in the case used in their specification.
Command line options will still be matched without case sensitivity.
This is useful for specifying option names in camel-case.

@item prog-desc @strong{and}
@itemx opts-ptr
@vindex prog-desc
@vindex opts-ptr
These define global pointer variables that point to the program
descriptor and the first option descriptor for a library option.  This
is intended for use by certain libraries that need command line and/or
initialization file option processing.  These definitions have no effect
on the option template output, but are used for creating a library
interface file.  Normally, the first "option" for a library will be a
documentation option that cannot be specified on the command line, but
is marked as @code{settable}.  The library client program will invoke the
@code{SET_OPTION} macro which will invoke a handler function that will
finally set these global variables.

@item usage
@vindex usage
Optionally names the usage procedure, if the library routine
@code{optionUsage()} does not work for you.  If you specify
@code{my_usage} as the value of this attribute, for example, you will
use a procedure by that name for displaying usage.  Of course, you will
need to provide that procedure and it must conform to this profile:
@example
void @i{my_usage}( tOptions* pOptions, int exitCode )
@end example

@item gnu-usage
@vindex gnu-usage
Normally, the default format produced by the @code{optionUsage} procedure
is @i{AutoOpts Standard}.  By specifying this attribute, the default format
will be @i{GNU-ish style}.  Either default may be overridden by the user with
the @code{AUTOOPTS_USAGE} environment variable.  If it is set to @code{gnu}
or @code{autoopts}, it will alter the style appropriately.  This attribute
will conflict with the @code{usage} attribute.

@item reorder-args
@vindex reorder-args
Some applications traditionally require that the command operands be
intermixed with the command options.  In order to handle that, the arguments
must be reordered.  If you are writing such an application, specify this
global option.  All of the options (and any associated option arguments)
will be brought to the beginning of the argument list.  New applications
should not use this feature, if at all possible.  This feature is
@i{disabled} if @code{POSIXLY_CORRECT} is defined in the environment.
@end table

@node Generated main
@subsection Generating main procedures
@cindex main procedure

When AutoOpts generates the code to parse the command line options, it has
the ability to produce any of several types of @code{main()} procedures.
This is done by specifying a global structured value for
@vindex main
@code{main}.  The values that it contains are dependent on the value set for
the one value it must have: @code{main-type}.

@vindex main-type
The recognized values for @code{main-type} are:
@menu
* main guile::              guile: main and inner_main procedures
* main shell-process::      shell-process: emit Bourne shell results
* main shell-parser::       shell-parser: emit Bourne shell script
* main main::               main: user supplied main procedure
* main include::            include: code emitted from included template
* main invoke::             invoke: code emitted from AutoGen macro
* main for-each::           for-each: perform function on each argument
@end menu

Here is an example of an @code{include} variation:

@example
main = @{
  main-type = include;
  tpl       = "main-template.tpl";
@};
@end example

@node main guile
@subsubsection guile: main and inner_main procedures

When the @code{main-type} is specified to be @code{guile},
a @code{main()} procedure is generated that calls @code{gh_enter()}, providing
it with a generated @code{inner_main()} to invoke.  If you must perform
certain tasks before calling @code{gh_enter()}, you may specify such code
in the value for the
@vindex before-guile-boot
@code{before-guile-boot} attribute.

The @code{inner_main()} procedure itself will process the command line
arguments (by calling @code{optionProcess()},
@pxref{libopts-optionProcess}), and then either invoke the code
specified with the
@vindex guile-main
@code{guile-main} attribute, or else export the parsed options to Guile
symbols and invoke the @code{scm_shell()} function from the Guile library.
This latter will render the program nearly identical to the stock
@code{guile(1)} program.

@node main shell-process
@subsubsection shell-process: emit Bourne shell results

This will produce a @code{main()} procedure that parses the command line
options and emits to @file{stdout} Bourne shell commands that puts the
option state into environment variables.  This can be used within a
shell script as follows:

@example
unset OPTION_CT
eval "`opt_parser \"$@@\"`"
test -z "$@{OPTION_CT@}" && exit 1
test $@{OPTION_CT@} -gt 0 && shift $@{OPTION_CT@}
@end example

If the option parsing code detects an error or a request for usage,
it will not emit an assignment to OPTION_CT and the script should just
exit.  If the options are set consistently, then something along the
lines of the following will be written to @file{stdout} and evaled:

@example
    OPTION_CT=4
    export OPTION_CT
    MYPROG_SECOND='first'
    export MYPROG_SECOND
    MYPROG_ANOTHER=1 # 0x1
    export MYPROG_ANOTHER
@end example

@noindent
If the arguments are to be reordered, however, then the resulting set
of operands will be emitted and @code{OPTION_CT} gets set to zero.
For example, the following would be appended to the above:

@example
    set -- 'operand1' 'operand2' 'operand3'
    OPTION_CT=0
@end example

@noindent
@code{OPTION_CT} is set to zero since it is not necessary to shift
off any options.

@node main shell-parser
@subsubsection shell-parser: emit Bourne shell script

This will produce a @code{main()} procedure that emits a shell script
that will parse the command line options.  That script can be emitted
to @file{stdout} or inserted or substituted into a pre-existing shell
script file.  Improbable markers are used to identify previously inserted
parsing text:

@example
# # # # # # # # # # -- do not modify this marker --
@end example

@noindent
The program is also pretty insistent upon starting its parsing script
on the second line.

@node main main
@subsubsection main: user supplied main procedure

You must supply a value for the @code{main-text} attribute.
You may also supply a value for
@vindex option-code
@code{option-code}.  If you do, then the @code{optionProcess} invocation
will not be emitted into the code.  AutoOpts will wrap the @code{main-text}
inside of:

@example
int
main( int argc, char** argv )
@{
    @{ // replaced by option-code, if that exists
        int ct = optionProcess( &<<prog-name>>Options, argc, argv );
        argc -= ct;
        argv += ct;
    @}
<<your main-text goes here>>
@}
@end example

@noindent
so you can most conveniently set the value with a ``@code{here string}''
(@pxref{here-string}):

@example
code = <<- _EndOfMainProc_
	<<your text goes here>>
	_EndOfMainProc_;
@end example

@node main include
@subsubsection include: code emitted from included template

You must write a template to produce your main procedure.
You specify the name of the template with the @code{tpl} attribute
and it will be incorporated at the point where AutoOpts is ready
to emit the @code{main()} procedure.

This can be very useful if, in your working environment, you have
many programs with highly similar @code{main()} procedures.  All you need
to do is parameterize the variations and specify which variant is needed
within the @code{main} AutoOpts specification.  Since you are coding
the template for this, the attributes needed for this variation would
be dictated by your template.

@node main invoke
@subsubsection invoke: code emitted from AutoGen macro

You must write a template to produce your main procedure.  That template
must contain a definition for the function specified with the @code{func}
attribute to this @code{main()} procedure specification.  Typically, this
template will be incorporated by using the @code{--lib-template} option
(@pxref{autogen lib-template}) in the AutoGen invocation.  Otherwise, this
variation operates in much the same way as ``@code{include}''
(@pxref{main include}) method.

@node main for-each
@subsubsection for-each: perform function on each argument

This produces a main procedure that invokes a procedure once for each operand
on the command line (non-option arguments), @strong{OR} once for each
non-blank, non-comment @code{stdin} input line.  Leading and trailing white
space is trimmed from the input line and comment lines are lines that are
empty or begin with a comment character, defaulting to a hash ('#') character.

@strong{NB}:
The @code{argument} program attribute (@pxref{program attributes})
must begin with the @code{[} character, to indicate that there are
command operands, but that they are optional.

There are a number of attributes to @code{main} that may be used:

@table @code
@item  handler-proc
@vindex handler-proc
This attribute is required.  It is used to name the procedure to call.
That procedure is presumed to be external, but if you provide the code
for it, then the procedure is emitted as a static procedure in the
generated code.

This procedure should return 0 on success, a cumulative error code on warning
and exit without returning on an unrecoverable error.  As the cumulative
warning codes are @i{or}-ed together, the codes should be some sort of bit
mask in order to be ultimately decipherable (if you need to do that).

If the called procedure needs to cause a fail-exit, it is expected to call
@code{exit(3)} directly.  If you want to cause a warning exit code, then this
handler function should return a non-zero status.  That value will be
@strong{OR}-ed into a result integer for computing the final exit code.  E.g.,
here is part of the emitted code:

@example
  int res = 0;
  if (argc > 0) @{
     do  @{
         res |= @i{my_handler}( *(argv++) );
     @} while (--argc > 0);
  @} else @{ ...
@end example

@item handler-type
@vindex handler-type
If you do not supply this attribute, your handler procedure must be
the default type.  The profile of the procedure must be:

@example
int @i{my_handler}( char const *pz_entry );
@end example

@noindent
However, if you do supply this attribute, you may set the value to any of
four alternate flavors:

@table @samp
@item name-of-file
This is essentially the same as the default handler type, except that before
your procedure is invoked, the generated code has verified that the string
names an existing file.  The profile is unchanged.

@item file-X
Before calling your procedure, the file is f-opened according to the ``X'',
where ``X'' may be any of the legal modes for @code{fopen(3C)}.  In this case,
the profile for your procedure must be:

@example
int @i{my_handler}( char const* pz_fname, FILE* entry_fp );
@end example

@item  text-of-file
@itemx some-text-of-file
Before calling your procedure, the contents of the file are read or mapped into memory.
(Excessively large files may cause problems.)  The @samp{some-text-of-file}
disallows empty files.  Both require regular files.  In this case, the profile
for your procedure must be:

@example
program_exit_code_t
@i{my_handler}(char const* pz_fname, char* file_text,
           size_t text_size);
@end example

@noindent
Note that though the @code{file_text} is not @code{const}, any changes made to
it are not written back to the original file.  It is merely a memory image of
the file contents.  Also, the memory allocated to hold the text is
@code{text_size + 1} bytes long and the final byte is always @code{NUL}.
The file contents need not be text, as the data are read with the @code{read(2)}
system call.
@end table

If you select one of these file type handlers, then on access or usage errors
the @code{PROGRAM_EXIT_FAILURE} exit code will, by default, be or-ed
into the final exit code.  This can be changed by specifying the
global @code{file-fail-code} attribute and naming a different value.
That is, something other than @code{failure}.  You may choose @code{success},
in which case file access issues will not affect the exit code and the error
message will not be printed.

@item @i{my_handler}-code
@vindex MYHANDLER-code
With this attribute, you provide the code for your handler procedure
in the option definition file.  In this case, your @code{main()}
procedure specification might look something like this:

@example
main = @{
  main-type    = for-each;
  handler-proc = @i{my_handler};
  @i{my_handler}-code = <<- EndOfMyCode
	/* whatever you want to do */
	EndOfMyCode;
@};
@end example

@noindent
and instead of an emitted external reference, a procedure will be emitted
that looks like this:

@example
static int
@i{my_handler}( char const* pz_entry )
@{
    int res = 0;
    <<@i{my_handler}-code goes here>>
    return res;
@}
@end example

@item main-init
@vindex main-init
This is code that gets inserted after the options have been processed, but
before the handler procs get invoked.

@item main-fini
@vindex main-fini
This is code that gets inserted after all the entries have been processed,
just before returning from @code{main()}.

@item comment-char
@vindex comment-char
If you wish comment lines to start with a character other than a hash
(@code{#}) character, then specify one character with this attribute.
If that character is the @code{NUL} byte, then only blank lines will be
considered comments.
@end table

@node option attributes
@subsection Option Attributes
@cindex option attributes

For each option you wish to specify, you must have a block macro named
@code{flag} defined.  There are two required attributes: @code{name} and
@code{descrip}.  If any options do not have a @code{value} (traditional flag
character) attribute, then the @code{long-opts} program attribute must also
be defined.  As a special exception, if no options have a @code{value}
@strong{and} @code{long-opts} is not defined @strong{and} @code{argument} is
not defined, then all arguments to the program are named options.  In this
case, the @code{-} and @code{--} command line option markers are optional.

@menu
* Required Attributes::         Required Attributes
* Common Attributes::           Common Option Attributes
* Immediate Action::            Immediate Action Attributes
* Option Conflict Attributes::  Option Conflict Attributes

These option attributes do not fit well with the above categories.

* opt-attr settable::           Program may set option
* opt-attr no-preset::          Option cannot be pre-configured
* opt-attr equivalence::        Option Equivalence Class
* opt-attr aliases::            Option Aliasing
* opt-attr default option::     Default Option
* opt-attr documentation::      Option Sectioning Comment
* opt-attr translators::        Translator Notes
@end menu

@node Required Attributes
@subsubsection Required Attributes
@cindex Required Attributes

Every option must have exactly one copy of both of these attributes.

@table @samp
@item name
@vindex name
Long name for the option.  Even if you are not accepting long options
and are only accepting flags, it must be provided.  AutoOpts generates
private, named storage that requires this name.  This name also causes
a @code{#define}-d name to be emitted.  It must not conflict with any
other names you may be using in your program.

For example, if your option name is, @code{debug} or @code{munged-up},
you must not use the @code{#define} names @code{DEBUG} (or
@code{MUNGED_UP}) in your program for non-AutoOpts related purposes.
They are now used by AutoOpts.

Sometimes (most especially under Windows), you may get a surprise.
For example, @code{INTERFACE} is apparently a user space name that
one should be free to use.  Windows usurps this name.  To solve this,
you must do one of the following:

@enumerate
@item
Change the name of your option
@item
add the program attribute (@pxref{program attributes}):

@example
export = '#undef INTERFACE';
@end example
@item
add the program attribute:

@example
guard-option-names;
@end example
@end enumerate

@item descrip
@vindex descrip
Except for documentation options, a @strong{very} brief description of the
option.  About 40 characters on one line, maximum, not counting any texinfo
markups.  Texinfo markups are stripped before printing in the usage text.  It
appears on the @code{usage()} output next to the option name.

If, however, the option is a documentation option, it will appear on one or
more lines by itself.  It is thus used to visually separate and comment upon
groups of options in the usage text.
@end table

@node Common Attributes
@subsubsection Common Option Attributes
@cindex Common Option Attributes

These option attributes are optional.  Any that do appear in the
definition of a flag, may appear only once.

@table @samp
@item value
@vindex value
The flag character to specify for traditional option flags, e.g., @code{-L}.

@item max
@vindex max
Maximum occurrence count (invalid if @var{disable} present).
The default maximum is 1.  @code{NOLIMIT} can be used for the value,
otherwise it must be a number or a @code{#define} that evaluates to a number.

@item min
@vindex min
Minimum occurrence count.  If present, then the option @strong{must}
appear on the command line.  Do not define it with the value zero (0).

@item must-set
@vindex must-set
If an option must be specified, but it need not be specified on
the command line, then specify this attribute for the option.

@item deprecated
@vindex deprecated
There are two effects to this attribute:  the usage text will not
show the option, and the generated documentation will mark it with:
``@emph{NOTE: THIS OPTION IS DEPRECATED}''.

@item disable
@vindex disable
Prefix for disabling (inverting sense of) the option.  Only useful
if long option names are being processed.  When an option has this
attribute, the test @code{ENABLED_OPT(OPTNAME)} is false when either
of the following is true:
@itemize @bullet
@item
The option has not been specified and the @code{enable} attribute has
not been specified.
@item
The option has been specified with this disabling prefix.
@end itemize
To detect that the option has been specified with the disabling
prefix, you must use:
@example
HAVE_OPT(OPTNAME) && ! ENABLED_OPT(OPTNAME)
@end example

@item enable
@vindex enable
Long-name prefix for enabling the option (invalid if @var{disable}
@strong{not} present).  Only useful if long option names are being
processed.

@item enabled
@vindex enabled
If default is for option being enabled.  (Otherwise, the OPTST_DISABLED
bit is set at compile time.)  Only useful if the option can be disabled.

@item ifdef
@itemx ifndef
@itemx omitted-usage
@vindex ifdef
@vindex ifndef
@vindex omitted-usage
If an option is relevant on certain platforms or when certain features
are enabled or disabled, you can specify the compile time flag used
to indicate when the option should be compiled in or out.  For example,
if you have a configurable feature, @code{mumble} that is indicated
with the compile time define, @code{WITH_MUMBLING}, then add:

@example
ifdef = WITH_MUMBLING;
@end example

@noindent
Take care when using these.  There are several caveats:

@itemize @bullet
@item
The case and spelling must match whatever is specified.
@item
Do not confuse these attributes with the AutoGen directives of the
same names, @xref{Directives}.  These cause C preprocessing directives
to be inserted into the generated C text.
@item
Only one of @code{ifdef} and @code{ifndef} may apply to any one option.
@item
The @code{VALUE_OPT_} values are @code{#define}-d.  If @code{WITH_MUMBLING}
is not defined, then the associated @code{VALUE_OPT_} value will not be
@code{#define}-d either.  So, if you have an option named, @code{MUMBLING}
that is active only if @code{WITH_MUMBLING} is @code{#define}-d, then
@code{VALUE_OPT_MUMBLING} will be @code{#define}-d iff @code{WITH_MUMBLING}
is @code{#define}-d.  Watch those switch statements.
@item
If you specify @code{omitted-usage}, then the option will be recognized
as disabled when it is configured out of the build, but will yield the
message, ``This option has been disabled.''  You may specify an alternate
message by giving @code{omitted-usage} a string value. e.g.:
@example
omitted-usage = 'you cannot do this';
@end example
@end itemize

@item no-command
@vindex no-command
This option specifies that the option is not allowed on the command line.
Such an option may not take a @code{value} (flag character) attribute.
The program must have the @code{homerc} (@pxref{program attributes}) option set.
@end table

@node Immediate Action
@subsubsection Immediate Action Attributes
@cindex immediate action

Certain options may need to be processed early.  For example, in order to
suppress the processing of configuration files, it is necessary to process the
command line option @code{--no-load-opts} @strong{before} the config files are
processed.  To accommodate this, certain options may have their enabled or
disabled forms marked for immediate processing.  The consequence of this is
that they are processed ahead of all other options in the reverse of normal
order.

Normally, the first options processed are the options specified in the first
@code{homerc} file, followed by then next @code{homerc} file through to the
end of config file processing.  Next, environment variables are processed and
finally, the command line options.  The later options override settings
processed earlier.  That actually gives them higher priority.  Command line
immediate action options actually have the lowest priority of all.  They would
be used only if they are to have an effect on the processing of subsequent
options.

@table @samp
@item immediate
@vindex immediate
Use this option attribute to specify that the enabled form of the option
is to be processed immediately.  The @code{help} and @code{more-help}
options are so specified.  They will also call @code{exit()} upon
completion, so they @strong{do} have an effect on the processing
of the remaining options :-).

@item immed-disable
@vindex immed-disable
Use this option attribute to specify that the disabled form of the
option is to be processed immediately.  The @code{load-opts} option is
so specified.  The @code{--no-load-opts} command line option will
suppress the processing of config files and environment variables.
Contrariwise, the @code{--load-opts} command line option is
processed normally.  That means that the options specified in that file
will be processed after all the @code{homerc} files and, in fact, after
options that precede it on the command line.

@item also
If either the @code{immediate} or the @code{immed-disable} attributes
are set to the string, ``@code{also}'', then the option will actually be
processed twice:  first at the immediate processing phase and again
at the ``normal'' time.
@end table

@node Option Conflict Attributes
@subsubsection Option Conflict Attributes
@cindex Option Conflict Attributes

These attributes may be used as many times as you need.
They are used at the end of the option processing to verify
that the context within which each option is found does not
conflict with the presence or absence of other options.

This is not a complete cover of all possible conflicts and
requirements, but it simple to implement and covers the
more common situations.

@table @samp
@cindex flags-must
@item flags-must
one entry for every option that @strong{must} be present
when this option is present

@cindex flags-cant
@item flags-cant
one entry for every option that @strong{cannot} be present
when this option is present
@end table

@node opt-attr settable
@subsubsection Program may set option
@vindex settable
If the option can be set outside of option processing, specify
``@code{settable}''.  If this attribute is defined, special macros for setting
this particular option will be inserted into the interface file.  For example,
@code{TEMPL_DIRS} is a settable option for AutoGen, so a macro named
@code{SET_OPT_TEMPL_DIRS(a)} appears in the interface file.  This attribute
interacts with the @var{documentation} attribute.

@node opt-attr no-preset
@subsubsection Option cannot be pre-configured
@vindex no-preset
@cindex configuration file
If presetting this option is not allowed, specify ``@code{no-preset}''.
(Thus, environment variables and values set in configuration files will be
ignored.)

@node opt-attr equivalence
@subsubsection Option Equivalence Class
@vindex equivalence
Generally, when several options are mutually exclusive and basically serve the
purpose of selecting one of several processing modes, specify the
``@code{equivalence}'' attribute.  These options will be considered an
equivalence class.  Sometimes, it is just easier to deal with them as such.
All members of the equivalence class must contain the same equivalenced-to
option, including the equivalenced-to option itself.  Thus, it must be a class
member.

For an option equivalence class, there is a single occurrence counter for
the class.  It can be referenced with the interface macro,
@code{COUNT_OPT(BASE_OPTION)}, where ``BASE_OPTION'' is the equivalenced-to
option name.

Also, please take careful note: since the options are mapped to the
equivalenced-to option descriptor, any option argument values are mapped to
that descriptor also.  Be sure you know which ``equivalent option'' was
selected before getting an option argument value!

During the presetting phase of option processing (@pxref{Presetting
Options}), equivalenced options may be specified.  However, if different
equivalenced members are specified, only the last instance will be
recognized and the others will be discarded.  A conflict error is indicated
only when multiple different members appear on the command line itself.

As an example of where equivalenced options might be useful, @code{cpio(1)}
has three options @code{-o}, @code{-i}, and @code{-p} that define the
operational mode of the program (@code{create}, @code{extract} and
@code{pass-through}, respectively).  They form an equivalence class from
which one and only one member must appear on the command line.  If
@code{cpio} were an AutoOpt-ed program, then each of these option
definitions would contain:

@example
equivalence = create;
@end example

and the program would be able to determine the operating mode
with code that worked something like this:

@example
switch (WHICH_IDX_CREATE) @{
case INDEX_OPT_CREATE:       ...
case INDEX_OPT_EXTRACT:      ...
case INDEX_OPT_PASS_THROUGH: ...
default:    /* cannot happen */
@}
@end example

@node opt-attr aliases
@subsubsection Option Aliasing

Sometimes, for backwards compatibility or tradition or just plain convenience,
it works better to define one option as a pure alias for another option.
For such situations, provide the following pieces of information:
@example
flag = @{
   name  = @i{aliasing-option-name};
   value = @i{aliasing-flag-char}; // optional !
   aliases = @i{aliased-to-option};
@};
@end example
Do not provide anything else.  The usage text for such an option will be:
@example
   This is an alias for @i{aliased-to-option}
@end example

@node opt-attr default option
@subsubsection Default Option
@vindex default
If your program processes its arguments in named option mode (See
@code{long-opts} in @ref{program attributes}), then you may select
@strong{one} of your options to be the default option.  Do so by using
attribute @code{default} with one of the options.  The option so specified
must have an @code{arg-type} (@pxref{Option Arguments}) specified, but not the
@code{arg-optional} (@pxref{arg-optional}) attribute.  That is to say, the
option argument must be required.

If you have done this, then any arguments that do not match an option name and
do not contain an equal sign (@code{=}) will be interpreted as an option
argument to the default option.

@node opt-attr documentation
@subsubsection Option Sectioning Comment
This attribute means the option exists for the purpose of separating option
description text in the usage output and texi documentation.  Without this
attribute, every option is a separate node in the texi docs.  With this
attribute, the documentation options become texi doc nodes and the options are
collected under them.  Choose the name attribute carefully because it will
appear in the texi documentation.

Libraries may also choose to make it settable so that the library can
determine which command line option is the first one that pertains to the
library.

@vindex documentation
If the @samp{documentation} attribute is present, then all other
attributes are disabled except @code{settable}, @code{call-proc} and
@code{flag-code}.  @code{settable} must be and is only specified if
@code{call-proc}, @code{extract-code} or @code{flag-code} has been specified.
When present, the @code{descrip} attribute will be displayed only when the
@code{--help} option has been specified.  It will be displayed flush to the
left hand margin and may consist of one or more lines of text, filled to 72
columns.

The name of the option will not be printed in the help text.  It @i{will},
however, be printed as section headers in the texi documentation.  If the
attribute is given a non-empty value, this text will be reproduced in the man
page and texi doc immediately after the @code{descrip} text.

@node opt-attr translators
@subsubsection Translator Notes
@vindex translators
If you need to give the translators a special note about a particular option,
please use the ``@code{translators}'' attribute.  The attribute text will be
emitted into the generated @code{.c} text where the option related strings get
defined.  To make a general comment about all of the option code, add comments
to an @code{include} attribute (@pxref{program attributes}).  Do @strong{not}
use this attribute globally, or it will get emitted into every option
definition block.

@node Option Arguments
@subsection Option Argument Specification
@cindex Option Arguments

Command line options come in three flavors:  options that do not
take arguments, those that do and those that may.  Without an
"arg-type" attribute, AutoOpts will not process an argument to an
option.  If "arg-type" is specified and "arg-optional" is also
specified, then the next command line token will be taken to
be an argument, unless it looks like the name of another option.

If the argument type is specified to be anything other than "str[ing]", then
AutoOpts will specify a callback procedure to handle the argument.  Some of
these procedures will be created and inserted into the generated @code{.c}
file, and others are already built into the @file{libopts} library.
Therefore, if you write your own callback procedure
(@pxref{Option Argument Handling}), then you must either not specify an
"arg-type" attribute, or else specify it to be of type "str[ing]".  Your
callback function will be able to place its own restrictions on what that
string may contain or represent.

Option argument handling attributes depend upon the value set for the
@vindex arg-type
@code{arg-type} attribute.  It specifies the type of argument the option
will take.  If not present, the option cannot take an argument.  If present,
it must be an entry in the following table.  The first three letters is
sufficient.

@menu
* arg-type string::         Arg Type String
* arg-type number::         Arg Type Number
* arg-type boolean::        Arg Type Boolean
* arg-type keyword::        Arg Type Keyword
* arg-type set membership:: Arg Type Set Membership
* arg-type hierarchy::      Arg Type Hierarchical
* arg-type file name::      Arg Type File Name
* arg-type time-duration::  Arg Type Time Duration
* arg-type time-date::      Arg Type Time and Date

Supporting attributes for particular argument types:

* arg-keyword::             Keyword list
* arg-optional::            Option Argument Optional
* arg-default::             Default Option Argument Value
@end menu

@node arg-type string
@subsubsection Arg Type String
@code{arg-type = string;}

The argument may be any arbitrary string, though your program or option
callback procedure may place additional constraints upon it.


@node arg-type number
@subsubsection Arg Type Number
@code{arg-type = number;}

The argument must be a correctly formed integer, without any trailing U's or
L's.  AutoOpts contains a library procedure to convert the string to a number.
If you specify range checking with @code{arg-range} (see below), then AutoOpts
produces a special purpose procedure for this option.

@table @samp
@item scaled
@vindex scaled
@code{scaled} marks the option so that suffixes of @samp{k}, @samp{K},
@samp{m}, @samp{M}, @samp{g}, @samp{G}, @samp{t}, and @samp{T} will multiply
the given number by a power of 1000 or 1024.  Lower case letters scale by a
power of 1000 and upper case scale by a power of 1024.

@item arg-range
@vindex arg-range
@code{arg-range} is used to create a callback procedure for validating the
range of the option argument.  It must match one of the range entries.  Each
@code{arg-range} should consist of either an integer by itself or an integer
range.  The integer range is specified by one or two integers separated by the
two character sequence, @code{->}.  Be sure to quote the entire range string.
The definitions parser will not accept the range syntax as a single string
token.

The generated procedure imposes the range constraints as follows:
@itemize @bullet
@item
A number by itself will match that one value.
@item
The high end of the range may not be @code{INT_MIN}, both for obvious
reasons and because that value is used to indicate a single-valued match.
@item
An omitted lower value implies a lower bound of INT_MIN.
@item
An omitted upper value implies a upper bound of INT_MAX.
@item
The argument value is required.  It may not be optional.
@item
The value must match one of the entries.  If it can match more than one,
then you have redundancies, but no harm will come of it.
@end itemize
@end table


@node arg-type boolean
@subsubsection Arg Type Boolean
@code{arg-type = boolean;}

The argument will be interpreted and always yield either AG_TRUE or
AG_FALSE.  False values are@:  the empty string, the number zero, or a
string that starts with @code{f}, @code{F}, @code{n} or @code{N}
(representing False or No).  Anything else will be interpreted as True.


@node arg-type keyword
@subsubsection Arg Type Keyword
@code{arg-type = keyword;}

The argument must match a specified list of strings (@pxref{arg-keyword}).
Assuming you have named the option, @code{optn-name}, the strings will be
converted into an enumeration of type @code{te_Optn_Name} with the values
@code{OPTN_NAME_KEYWORD}.*  If you have @strong{not} specified a default value,
the value @code{OPTN_NAME_UNDEFINED} will be inserted with the value zero.
The option will be initialized to that value.  You may now use this in your
code as follows:

@example
te_Optn_Name opt = OPT_VALUE_OPTN_NAME;
switch (opt) @{
case OPTN_NAME_UNDEFINED:  /* undefined things */ break;
case OPTN_NAME_KEYWORD:    /* `keyword' things */ break;
default: /* utterly impossible */ ;
@}
@end example

AutoOpts produces a special purpose procedure for this option.
You may not specify an alternate handling procedure.

If you have need for the string name of the selected keyword, you
may obtain this with the macro, @code{OPT_OPTN_NAME_VAL2STR(val)}.
The value you pass would normally be @code{OPT_VALUE_OPTN_NAME},
but anything with numeric value that is legal for @code{te_Optn_Name}
may be passed.  Anything out of range will result in the string,
@code{"*INVALID*"} being returned.  The strings are read only.
It may be used as in:

@example
te_Optn_Name opt = OPT_VALUE_OPTN_NAME;
printf( "you selected the %s keyword\n",
        OPT_OPTN_NAME_VAL2STR(opt) );
@end example

* Note: you may replace the @code{OPTN_NAME} enumeration prefix with
another prefix by specifying a
@vindex prefix-enum
@code{prefix-enum} attribute.

Finally, users may specify the argument either by name or by number.
Since the numeric equivalents change by having new entries inserted
into the keyword list, this would not be a recommended practice.
However, either @code{-1} or @code{~0} will always be equivalent to
specifying the last keyword.

@node arg-type set membership
@subsubsection Arg Type Set Membership
@code{arg-type = set;}

The argument must be a list of names each of which must match the strings
"@code{all}", "@code{none}" or one of the keywords (@pxref{arg-keyword})
specified for this option.  @code{all} will turn on all membership bits and
@code{none} will turn them all off.  Specifying one of the keywords will turn
on the corresponding set membership bit.  Literal numbers may also be used and
may, thereby, set or clear more than one bit.  Preceding a keyword or literal
number with a bang (@code{!}  - exclamation point) will turn the bit(s) off.
The number of keywords allowed is constrained by the number of bits in a
pointer, as the bit set is kept in a @code{void*}.

If, for example, you specified @code{first} in your list of keywords,
then you can use the following code to test to see if either @code{first}
or @code{all} was specified:

@example
uintptr_t opt = OPT_VALUE_OPTN_NAME;
if (opt & OPTN_NAME_FIRST)
    /* OPTN_NAME_FIRST bit was set */ ;
@end example

AutoOpts produces a special purpose procedure for this option.
To set multiple bits as the default (initial) value, you must
specify an initial numeric value (which might become inaccurate over
time), or else specify @code{arg-default} multiple times.  Do not
specify a series of names conjoined with @code{+} symbols as the
value for any of the @code{arg-default} attributes.  That works for
option parsing, but not for the option code generation.

@node arg-type hierarchy
@subsubsection Arg Type Hierarchical
@code{arg-type = hierarchy;}
@*
@code{arg-type = nested;}

This denotes an option with a structure-valued argument, a.k.a.
``subopts'' in @code{getopts} terminology.  The argument is parsed
and the values made available to the program via the find and
find next calls (@xref{libopts-optionFindValue},
@xref{libopts-optionGetValue}, and
@pxref{libopts-optionFindNextValue}).

@example
tOptionValue * val = optionGetValue(VALUE_OPT_OPTN_NAME, "name");
while (val != NULL) @{
  process(val);
  val = optionNextValue(VALUE_OPT_OPTN_NAME, val);
  if (wrong_name(val, "name"))
    break;
@}
@end example


@node arg-type file name
@subsubsection Arg Type File Name
@code{arg-type = file;}

This argument type will have some validations on the argument and,
optionally, actually open the file.  You must specify several additonal
attributes for the option:

@table @samp
@item file-exists
@vindex file-exists
If not specified or empty, then the directory portion of the name is checked.
The directory must exist or the argument is rejected and the usage procedure
is invoked.

Otherwise, both the directory as above and the full name is tested for
existence.  If the value begins with the two letters ``no'', then the file
must not pre-exist.  Otherwise, the file is expected to exist.

@item open-file
@vindex open-file
If not specified or empty, the file is left alone.
If the value begins with the four letters ``desc''[@i{riptor}], then
@code{open(2)} is used and @code{optArg.argFd} is set.  Otherwise, the
file is opened with @code{fopen} and @code{optArg.argFp} is set.

@item file-mode
@vindex file-mode
If ``open-file'' is set and not empty, then you must specify the open mode.
Set the value to the flag bits or mode string as appropriate for the open
type.
@end table


@node arg-type time-duration
@subsubsection Arg Type Time Duration
@code{arg-type = time-duration;}

The argument will be converted into a number of seconds.  It may be
a multi-part number with different parts being multiplied into a seconds
value and added into the final result.  Valid forms are in the table
below.  Upper cased letters represent numbers that must be used in the
expressions.

@table @samp
@item [[HH:]MM:]SS
@code{HH} is multiplied by @code{3600} and @code{MM} multiplied by @code{60}
before they are added to @code{SS}.  This time specification may not be
followed by any other time specs.  @code{HH} and @code{MM} are both optional,
though @code{HH} cannot be specified without @code{MM}.

@item DAYS d
@code{DAYS} is multiplied by the number of seconds in a day.  This value may
be followed by (and added to) values specified by @code{HH:MM:SS} or the
suffixed values below.  If present, it must always be first.

@item HRS h
@code{HRS} is multiplied by the number of seconds in an hour.  This value may
be followed by (and added to) values specified by @code{MM:SS} or the
suffixed values below.

@item MINS m
@code{MINS} is multiplied by the number of seconds in a minute.  This value may
be followed by (and added to) a count of seconds.

@item SECS s
This value can only be the last value in a time specification.  The @code{s}
suffix is optional.
@end table

@example
   5 d 1:10:05    ==> 5 days + 1 hour 10 minutes and 5 seconds
   5 d 1 h 10 m 5 ==> yields: 436205 seconds
   5d1h10m5s      ==> same result -- spaces are optional.
@end example

When saved into a config file, the value will be stored as a simple count
of seconds.  There are actually more (many) accepted time duration strings.
The full documentation can be found with ISO-8601 documentation and the
more extedded documentation when ``parse_duration()'' becomes more widely
available.


@node arg-type time-date
@subsubsection Arg Type Time and Date
@code{arg-type = time-date;}

The argument will be converted into the number of seconds since the epoch.
The conversion rules are very complicated, please see the @file{getdate_r(3GNU)}
man page.  There are some additional restrictions:

@enumerate
@item
Your project must be compiled with @code{PKGDATADIR} defined and naming a
valid directory.
@item
The @code{DATEMSK} environment variable will be set to the @file{datemsk} file
within that directory.
@end enumerate

If that file is not accessible for any reason, the string will be
parsed as a time duration (@pxref{arg-type time-duration}) instead of a
specific date and time.

@node arg-keyword
@subsubsection Keyword list
@vindex keyword
If the @code{arg-type} is @code{keyword} (@pxref{arg-type keyword}) or
@code{set-membership} (@pxref{arg-type set membership}), then you must specify
the list of keywords by a series of @code{keyword} entries.  The interface
file will contain values for @code{@i{<OPTN_NAME>}_@i{<KEYWORD>}} for each
keyword entry.  @code{keyword} option types will have an enumeration and
@code{set-membership} option types will have a set of unsigned bits
@code{#define}-d.

If the @code{arg-type} is specifically @code{keyword}, you may also add
special handling code with a
@vindex extra-code
@code{extra-code} attribute.  After @code{optionEnumerationVal} has
converted the input string into an enumeration, you may insert code to
process this enumeration value (@code{pOptDesc->optArg.argEnum}).

@node arg-optional
@subsubsection Option Argument Optional
@vindex arg-optional
This attribute indicates that the user does not have to supply an argument for
the option.  This is only valid if the @var{arg-type} is @code{string}
(@pxref{arg-type string}) or @code{keyword} (@pxref{arg-type keyword}).  If it
is @code{keyword}, then this attribute may also specify the default keyword to
assume when the argument is not supplied.  If left empty, @var{arg-default}
(@pxref{arg-default}) or the zero-valued keyword will be used.

This is overridden and the options are required if the libopts library
gets configured with @code{--disable-optional-args}.

@node arg-default
@subsubsection Default Option Argument Value
@vindex arg-default
This specifies the default option argument value to be used when the option is
not specified or preset.  You may specify multiple @code{arg-default} values if
the argument type is @code{set membership}.

@node Option Argument Handling
@subsection Option Argument Handling
@cindex Option Argument Handling

AutoOpts will either specify or automatically generate callback procedures
for options that take specialized arguments.  The only option argument types
that are not specialized are plain string arguments and no argument at all.
For options that fall into one of those two categories, you may specify your
own callback function, as specified below.  If you do this and if you
specify that options are resettable (@pxref{automatic options}), then your
option handling code @strong{must} look for the @samp{OPTST_RESET} bit in
the @code{fOptState} field of the option descriptor.

If the option takes a string argument, then you may specify that the option
is to be handled by the @code{libopts} library procedures
@code{stackOptArg()} or @code{unstackOptArg()} (see below).  In this case,
you may not provide option handling code.

Finally, @samp{documentation} options (@pxref{opt-attr documentation}) may
also be marked as @option{settable} (@pxref{opt-attr settable}) and have
special callback functions (either @samp{flag-code}, @samp{extract-code},
or @samp{call-proc}).

@table @samp
@item flag-code
@vindex flag-code
statements to execute when the option is encountered.  This may be used in
conjunction with option argument types that cause AutoOpts to emit handler
code.  If you do this, the @samp{flag-code} with index zero (0) is emitted
into the handler code @emph{before} the argument is handled, and the entry
with index one (1) is handled afterward.

The generated procedure will be laid out something like this:

@example
static void
doOpt<name>(tOptions* pOptions, tOptDesc* pOptDesc)
@{
<flag-code[0]>
<AutoOpts defined handler code>
<flag-code[1]>
@}
@end example

Only certain fields within the @code{tOptions} and @code{tOptDesc}
structures may be accessed.  @xref{Option Processing Data}.  When writing
this code, you must be very careful with the @code{pOptions} pointer.  The
handler code is called with this pointer set to special values for handling
special situations.  Your code must handle them.  As an example,
look at @code{optionEnumerationVal} in @file{enum.c}.

@item extract-code
@vindex extract-code
This is effectively identical to @code{flag-code}, except that the
source is kept in the output file instead of the definitions file
and you cannot use this in conjunction with options with arguments,
other than string arguments.

A long comment is used to demarcate the code.  You must not modify
that marker.  @i{Before} regenerating the option code file,
the old file is renamed from MUMBLE.c to MUMBLE.c.save.  The template
will be looking there for the text to copy into the new output file.

@item call-proc
@vindex call-proc
external procedure to call when option is encountered.  The calling
sequence must conform to the sequence defined above for the generated
procedure, @code{doOpt<name>}.  It has the same restrictions
regarding the fields within the structures passed in as arguments.
@xref{Option Processing Data}.

@item flag-proc
@vindex flag-proc
Name of another option whose @code{flag-code} can be executed
when this option is encountered.

@item stack-arg
@vindex stack-arg
Call a special library routine to stack the option's arguments.  Special
macros in the interface file are provided for determining how many of the
options were found (@code{STACKCT_OPT(NAME)}) and to obtain a pointer to a
list of pointers to the argument values (@code{STACKLST_OPT(NAME)}).
Obviously, for a stackable argument, the @code{max} attribute
(@pxref{Common Attributes}) needs to be set higher than @code{1}.

If this stacked argument option has a disablement prefix, then the entire
stack of arguments will be cleared by specifying the option with that
disablement prefix.

@item unstack-arg
@vindex unstack-arg
Call a special library routine to remove (``unstack'') strings
from a @code{stack-arg} option stack.  This attribute must name
the option that is to be ``unstacked''.  Neither this option nor
the stacked argument option it references may be equivalenced to
another option.
@end table

@node Internationalizing Options
@subsection Internationalizing Options
@cindex Internationalizing Options

Normally, AutoOpts produces usage text that is difficult to translate.  It is
pieced together on the fly using words and phrases scattered around here and
there, piecing together toe document.  This does not translate well.

Incorporated into this package are some ways around the problem.  First, you
should specify the @code{full-usage} and @code{short-usage} program attributes
(@pxref{program attributes}).  This will enable your translators to translate
the usage text as a whole.

Your translators will also be able to translate long option names.  The option
name translations will then become the names searched for both on the command
line and in configuration files.  However, it will not affect the names of
environment variable names used to configure your program.

If it is considered desireable to keep configuration files in the ``C''
locale, then several macros are available to suppress or delay the
translations of option names at run time.  These are all disabled if
@code{ENABLE_NLS} is not defined at compile time or if @code{no-xlate} has
been set to the value @emph{anything}.  These macros @strong{must}
be invoked before the first invocation of @code{optionProcess}.

@table @samp
@item  OPT_NO_XLAT_CFG_NAMES;
@itemx OPT_XLAT_CFG_NAMES;
Disable (or enable) the translations of option names for configuration files.
If you enable translation for config files, then they will be translated for
command line options.

@item  OPT_NO_XLAT_OPT_NAMES;
@itemx OPT_XLAT_OPT_NAMES;
Disable (or enable) the translations of option names for command line
processing.  If you disable the translation for command line processing,
you will also disable it for configuration file processing.  Once translated,
the option names will remain translated.
@end table

@node documentation attributes
@subsection Man and Info doc Attributes
@cindex documentation attributes

AutoOpts includes AutoGen templates for producing abbreviated man pages
and for producing the invoking section of an info document.  To take
advantage of these templates, you must add several attributes to your
option definitions.

@table @samp
@item arg-name
@vindex arg-name
If an option has an argument, the argument should have a name for
documentation purposes.  It will default to @code{arg-type}, but
it will likely be clearer with something else like, @code{file-name}
instead of @code{string} (the type).

@item doc
@vindex doc
First, every @code{flag} definition @emph{other than} ``documentation''
definitions, must have a @code{doc} attribute defined.  If the option takes
an argument, then it will need an @code{arg-name} attribute as well.  The
@code{doc} text should be in plain sentences with minimal formatting.  The
Texinfo commands @code{@@code}, and @code{@@var} will have its enclosed text
made into @strong{\fB} entries in the man page, and the @code{@@file} text
will be made into @strong{\fI} entries.  The @code{arg-name} attribute is
used to display the option's argument in the man page.

Options marked with the ``documentation'' attribute are for documenting
the usage text.  All other options should have the ``doc'' attribute in
order to document the usage of the option in the generated man pages.

@item option-info
@vindex option-info
This text will be inserted as a lead-in paragraph in the @code{OPTIONS}
section of the generated man page.

@item doc-section
@vindex doc-section
This is a compound attribute that requires three @i{sub}attributes:
@table @i
@item ds-type
This describes the section type.  Basically, the title of the section
that will be added to all output documentation.  There may be only one
@code{doc-section} for any given @code{ds-type}.  If there are duplicates,
the results are undefined (it might work, it might not).

There are five categories of @code{ds-type} sections.
They are those that the documentation templates would otherwise:
@enumerate
@item
always create itself, ignoring any @code{ds-type}s by this name.
These are marked, below, as @code{ao-only}.
@item
create, if none have been provided.
These are marked, @code{alternate}.
@item
create, but augment if the @code{doc-section} was provided.
These are marked, @code{augments}.
@item
do nothing, but inserts them into the output in a prescribed order.
These are marked, @code{known}
@item
knows nothing about them.  They will be alphabetized and inserted
after the list of leading sections and before the list of trailing
sections.  These are not marked because I don't know their names.
@end enumerate

Some of these are emitted by the documentation templates only if
certain conditions are met.  If there are conditions, they are
explained below.  If there are no conditions, then you will always
see the named section in the output.

The output sections will appear in this order:
@table @samp
@item NAME
@code{ao-only}.
@item SYNOPSIS
@code{alternate}.
@item DESCRIPTION
@code{augments}.
@item OPTIONS
@code{ao-only}.
@item OPTION PRESETS
@code{ao-only}, if environment presets or configuration file processing
has been specified.
@item unknown
At this point, the unknown, alphabetized sections are inserted.
@item IMPLEMENTATION NOTES
@code{known}
@item ENVIRONMENT
@code{augments}, if environment presets have been specified.
@item FILES
@code{augments}, if configuration file processing has been specified.
@item EXAMPLES
@code{known}
@item EXIT STATUS
@code{augments}.
@item ERRORS
@code{known}
@item COMPATIBILITY
@code{known}
@item SEE ALSO
@code{known}
@item CONFORMING TO
@code{known}
@item HISTORY
@code{known}
@item AUTHORS
@code{alternate}, if the @code{copyright} stanza has either
an @code{author} or an @code{owner} attribute.
@item COPYRIGHT
@code{alternate}, if there is a @code{copyright} stanza.
@item BUGS
@code{augments}, if the @code{copyright} stanza has an
@code{eaddr} attribute.
@item NOTES
@code{augments}.
@end table

@item ds-format
This describes the format of the associated @code{ds-text} section.
@code{man}, @code{mdoc} and @code{texi} formats are supported.
Regardless of the chosen format, the formatting tags in the output
text will be converted to @code{man} macros for @code{man} pages,
@code{mdoc} macros for @code{mdoc} pages, and @code{texi} macros for
@code{texinfo} pages.
@item ds-text
This is the descriptive text, written according to the rules for
@code{ds-format} documents.
@end table

Here is an example of a ``doc-section'' for a ``SEE ALSO'' type.

@example
doc-section = @{
  ds-type   = 'SEE ALSO'; // or anything else
  ds-format = 'man';      // or texi or mdoc format
  ds-text   = <<-_EOText_
	text relevant to this section type,
	in the chosen format
	_EOText_;
@};
@end example

@item prog-man-descrip
@itemx prog-info-descrip
@vindex prog-man-descrip
@vindex prog-info-descrip
These attributes are now deprecated.
Please use a @code{doc-section} stanza with a @code{ds-type}
attribute set to @code{DESCRIPTION} instead.

@item detail
@vindex detail
This attribute is used to add a very short explanation about what
a program is used for when the ``title'' attribute is insufficient.
If there is no ``doc-section'' stanza of type ``DESCRIPTION'', then
this text is used for the man page DESCRIPTION section, too.
@end table

@node automatic options
@subsection Automatically Supported Options
@cindex automatic options

AutoOpts provides automated support for several options.  @code{help} and
@code{more-help} are always provided.  The others are conditional upon
various global program attributes being defined @xref{program attributes}.

Below are the option names and default flag values.  The flags are activated
if and only if at least one user-defined option also uses a flag value.  The
long names are supported as option names if @code{long-opts} has been
specified.  These option flags may be deleted or changed to characters of your
choosing by specifying
@vindex more-help-value
@vindex usage-value
@vindex version-value
@vindex load-opts-value
@vindex reset-value
@code{xxx-value = "y";}, where @code{xxx} is one of the
option names below and @code{y} is either empty or the character of your choice.
For example, to change the help flag from @code{?} to @code{h}, specify
@vindex help-value
@code{help-value = "h";}; and to require that @code{save-opts} be specified
only with its long option name, specify
@vindex save-opts-value
@code{save-opts-value = "";}.

Additionally, the procedure that prints out the program version may be
replaced by specifying @code{version-proc}.
@vindex version-proc
This procedure must be defined to be of external scope (non-static).
By default, the AutoOpts library provides @code{optionPrintVersion}
and it will be the specified callback function in the option
definition structure.

With the exception of the @code{load-opts} option, none of these automatically
supported options will be recognized in configuration files or environment
variables.

@table @samp
@item help -?
This option will immediately invoke the @code{USAGE()} procedure
and display the usage line, a description of each option with
its description and option usage information.  This is followed
by the contents of the definition of the @code{detail} text macro.

@item more-help -!
This option is identical to the @code{help} option, except that the
output is passed through a pager program.  (@code{more} by default, or
the program identified by the @code{PAGER} environment variable.)

@item usage -u
This option must be requested by specifying, @code{usage-opt} in the option
definition file.  It will produce abbreviated help text to @file{stdout} and
exit with zero status (@code{EXIT_SUCCESS}).

@item version -v
This will print the program name, title and version.  If it is followed by
the letter @code{c} and a value for @code{copyright} and @code{owner} have
been provided, then the copyright will be printed, too.  If it is followed
by the letter @code{n}, then the full copyright notice (if available) will
be printed.  The @code{version} attribute must be specified in the option
definition file.

@item load-opts -<
@cindex configuration file
This option will load options from the named file.  They will be treated
exactly as if they were loaded from the normally found configuration files,
but will not be loaded until the option is actually processed.  This can also
be used within another configuration file, causing them to nest.  This is the
@strong{only} automatically supported option that can be activated inside of
config files or with environment variables.

Specifying the negated form of the option (@code{--no-load-opts}) will
suppress the processing of configuration files and environment variables.

This option is activated by specifying one or more @code{homerc} attributes.

@item save-opts ->
@cindex configuration file
This option will cause the option state to be printed in the configuration file
format when option processing is done but not yet verified for consistency.
The program will terminate successfully without running when this has
completed.  Note that for most shells you will have to quote or escape the
flag character to restrict special meanings to the shell.

The output file will be the configuration file name (default or provided by
@code{rcfile}) in the last directory named in a @code{homerc} definition.

This option may be set from within your program by invoking the
"@code{SET_OPT_SAVE_OPTS(@i{filename})}" macro (@pxref{SET_OPT_name}).
Invoking this macro will set the file name for saving the option processing
state, but the state will @strong{not} actually be saved.  You must call
@code{optionSaveFile} to do that (@pxref{libopts-optionSaveFile}).
@strong{CAVEAT:} if, after invoking this macro, you call
@code{optionProcess}, the option processing state will be saved to this file
and @code{optionProcess} will not return.  You may wish to invoke
@code{CLEAR_OPT( SAVE_OPTS )} (@pxref{CLEAR_OPT}) beforehand if you do need
to reinvoke @code{optionProcess}.

This option is activated by specifying one or more @code{homerc} attributes.

@item reset-option -R
This option takes the name of an option for the current program and resets its
state such that it is set back to its original, compile-time initialized
value.  If the option state is subsequently stored (via @code{--save-opts}),
the named option will not appear in that file.

This option is activated by specifying the @code{resettable} attribute.

@strong{BEWARE}:  If the @code{resettable} attribute is specified, all
option callbacks @strong{must} look for the @code{OPTST_RESET} bit in the
@code{fOptState} field of the option descriptor.  If set, the @code{optCookie}
and @code{optArg} fields will be unchanged from their last setting.  When the
callback returns, these fields will be set to their original values.  If you
use this feature and you have allocated data hanging off of the cookie, you
need to deallocate it.
@end table

@node standard options
@subsection Library of Standard Options
@cindex standard options

AutoOpts has developed a set of standardized options.
You may incorporate these options in your program simply by @emph{first}
adding a @code{#define} for the options you want, and then the line,

@example
#include stdoptions.def
@end example

@noindent
in your option definitions.  The supported options are specified thus:

@example
#define DEBUG
#define DIRECTORY
#define DRY_RUN
#define INPUT
#define INTERACTIVE
#define OUTPUT
#define WARN

#define SILENT
#define QUIET
#define BRIEF
#define VERBOSE
@end example

By default, only the long form of the option will be available.
To specify the short (flag) form, suffix these names with @code{_FLAG}.
e.g.,

@example
#define DEBUG_FLAG
@end example

@code{--silent}, @code{--quiet}, @code{--brief} and @code{--verbose} are
related in that they all indicate some level of diagnostic output.
These options are all designed to conflict with each other.
Instead of four different options, however, several levels can be
incorporated by @code{#define}-ing @code{VERBOSE_ENUM}.  In conjunction
with @code{VERBOSE}, it incorporates the notion of @i{5} levels in an
enumeration: @code{silent}, @code{quiet}, @code{brief},
@code{informative} and @code{verbose}; with the default being
@code{brief}.

@ignore
END   == AUTOOPTS-MAIN == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == AUTOOPTS-API == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c === SECTION MARKER

@node AutoOpts API
@section Programmatic Interface
@cindex AutoOpts API

The user interface for access to the argument information is completely
defined in the generated header file and in the portions of the
distributed file "options.h" that are marked "public".

In the following macros, text marked @code{<NAME>} or @code{name}
is the name of the option @strong{in upper case} and @strong{segmented
with underscores @code{_}}.  The macros and enumerations defined in the
options header (interface) file are used as follows:

To see how these @code{#define} macros are used in a program,
the reader is referred to the several @file{opts.h} files
included with the AutoGen sources.

@menu
* Option Processing Data::  Data for Option Processing
* CLEAR_OPT::               CLEAR_OPT( <NAME> ) - Clear Option Markings
* COUNT_OPT::               COUNT_OPT( <NAME> ) - Definition Count
* DESC::                    DESC( <NAME> ) - Option Descriptor
* DISABLE_OPT_name::        DISABLE_OPT_name - Disable an option
* ENABLED_OPT::             ENABLED_OPT( <NAME> ) - Is Option Enabled?
* ERRSKIP_OPTERR::          ERRSKIP_OPTERR - Ignore Option Errors
* ERRSTOP_OPTERR::          ERRSTOP_OPTERR - Stop on Errors
* HAVE_OPT::                HAVE_OPT( <NAME> ) - Have this option?
* ISSEL_OPT::               ISSEL_OPT( <NAME> ) - Is Option Selected?
* ISUNUSED_OPT::            ISUNUSED_OPT( <NAME> ) - Never Specified?
* OPTION_CT::               OPTION_CT - Full Count of Options
* OPT_ARG::                 OPT_ARG( <NAME> ) - Option Argument String
* OPT_NO_XLAT_CFG_NAMES::   OPT_NO_XLAT_CFG_NAMES - option name xlation
* OPT_NO_XLAT_OPT_NAMES::   OPT_NO_XLAT_OPT_NAMES - option name xlation
* OPT_VALUE_name::          OPT_VALUE_name - Option Argument Value
* OPT_XLAT_CFG_NAMES::      OPT_XLAT_CFG_NAMES - option name xlation
* OPT_XLAT_OPT_NAMES::      OPT_XLAT_OPT_NAMES - option name xlation
* RESTART_OPT::             RESTART_OPT( n ) - Resume Option Processing
* SET_OPT_name::            SET_OPT_name - Force an option to be set
* STACKCT_OPT::             STACKCT_OPT( <NAME> ) - Stacked Arg Count
* STACKLST_OPT::            STACKLST_OPT( <NAME> ) - Argument Stack
* START_OPT::               START_OPT - Restart Option Processing
* STATE_OPT::               STATE_OPT( <NAME> ) - Option State
* USAGE::                   USAGE( exit-code ) - Usage invocation macro
* VALUE_OPT_name::          VALUE_OPT_name - Option Flag Value
* VERSION::                 VERSION - Version and Full Version
* WHICH_IDX_name::          WHICH_IDX_name - Which Equivalenced Index
* WHICH_OPT_name::          WHICH_OPT_name - Which Equivalenced Option
* teOptIndex::              teOptIndex - Option Index and Enumeration
* OPTIONS_STRUCT_VERSION::  OPTIONS_STRUCT_VERSION - active version
* libopts procedures::      libopts External Procedures
@end menu

@node Option Processing Data
@subsection Data for Option Processing
@cindex Option Processing Data

This section describes the data that may be accessed from within the
option processing callback routines.  The following fields may be used
in the following ways and may be used for read only.  The first set is
addressed from the @code{tOptDesc*} pointer:

@table @samp
@cindex optIndex
@item optIndex
@cindex optValue
@item optValue
These may be used by option procedures to determine which option they
are working on (in case they handle several options).

@cindex optActualIndex
@item optActualIndex
@cindex optActualValue
@item optActualValue
These may be used by option procedures to determine which option was
used to set the current option.  This may be different from the above if
the options are members of an equivalence class.

@cindex optOccCt
@item optOccCt
If AutoOpts is processing command line arguments, then this value will
contain the current occurrence count.  During the option preset phase
(reading configuration files and examining environment variables), the value is
zero.

@cindex fOptState
@item fOptState
The field may be tested for the following bit values
(prefix each name with @code{OPTST_}, e.g. @code{OPTST_INIT}):

@table @samp
@item INIT
Initial compiled value.  As a bit test, it will always yield FALSE.

@item SET
The option was set via the @code{SET_OPT()} macro.

@item PRESET
@cindex configuration file
The option was set via a configuration file.

@item DEFINED
The option was set via a command line option.

@item SET_MASK
This is a mask of flags that show the set state, one of the
above four values.

@item EQUIVALENCE
This bit is set when the option was selected by an equivalenced option.

@item DISABLED
This bit is set if the option is to be disabled.
(Meaning it was a long option prefixed by the disablement prefix, or
the option has not been specified yet and initializes as @code{disabled}.)
@end table

As an example of how this might be used, in AutoGen I want to allow
template writers to specify that the template output can be left
in a writable or read-only state.  To support this, there is a Guile
function named @code{set-writable} (@pxref{SCM set-writable}).
Also, I provide for command options @code{--writable} and
@code{--not-writable}.  I give precedence to command line and RC
file options, thus:

@example
switch (STATE_OPT( WRITABLE )) @{
case OPTST_DEFINED:
case OPTST_PRESET:
    fprintf(stderr, zOverrideWarn, pCurTemplate->pzFileName,
            pCurMacro->lineNo);
    break;

default:
    if (gh_boolean_p( set ) && (set == SCM_BOOL_F))
        CLEAR_OPT( WRITABLE );
    else
        SET_OPT_WRITABLE;
@}
@end example

@cindex pzLastArg
@item pzLastArg
Pointer to the latest argument string.  BEWARE@: If the argument type
is numeric, an enumeration or a bit mask, then this will be the
argument @strong{value} and not a pointer to a string.
@end table

The following two fields are addressed from the @code{tOptions*} pointer:

@table @samp
@cindex pzProgName
@item pzProgName
Points to a NUL-terminated string containing the current program
name, as retrieved from the argument vector.

@cindex pzProgPath
@item pzProgPath
Points to a NUL-terminated string containing the full path of
the current program, as retrieved from the argument vector.
(If available on your system.)

@end table

Note@:  these fields get filled in during the first call to
@code{optionProcess()}.  All other fields are private, for the exclusive
use of AutoOpts code and are subject to change.

@node CLEAR_OPT
@subsection CLEAR_OPT( <NAME> ) - Clear Option Markings
@findex CLEAR_OPT

Make as if the option had never been specified.
@code{HAVE_OPT(<NAME>)} will yield @code{FALSE}
after invoking this macro.

@node COUNT_OPT
@subsection COUNT_OPT( <NAME> ) - Definition Count
@findex COUNT_OPT

This macro will tell you how many times the option was
specified on the command line.  It does not include counts
of preset options.

@example
if (COUNT_OPT( NAME ) != desired-count) @{
    make-an-undesirable-message.
@}
@end example

@node DESC
@subsection DESC( <NAME> ) - Option Descriptor
@findex DESC

This macro is used internally by other AutoOpt macros.
It is not for general use.  It is used to obtain the option description
corresponding to its @strong{UPPER CASED} option name argument.
This is primarily used in other macro definitions.

@node DISABLE_OPT_name
@subsection DISABLE_OPT_name - Disable an option
@findex DISABLE_OPT_name

This macro is emitted if it is both settable
and it can be disabled.  If it cannot be disabled, it may
always be CLEAR-ed (see above).

The form of the macro will actually depend on whether the
option is equivalenced to another, and/or has an assigned
handler procedure.  Unlike the @code{SET_OPT} macro,
this macro does not allow an option argument.

@example
DISABLE_OPT_NAME;
@end example

@node ENABLED_OPT
@subsection ENABLED_OPT( <NAME> ) - Is Option Enabled?
@findex ENABLED_OPT

Yields true if the option defaults to disabled and
@code{ISUNUSED_OPT()} would yield true.  It also yields true if
the option has been specified with a disablement prefix,
disablement value or the @code{DISABLE_OPT_NAME} macro was invoked.

@node ERRSKIP_OPTERR
@subsection ERRSKIP_OPTERR - Ignore Option Errors
@findex ERRSKIP_OPTERR

When it is necessary to continue (return to caller)
on option errors, invoke this option.  It is reversible.
@xref{ERRSTOP_OPTERR}.

@node ERRSTOP_OPTERR
@subsection ERRSTOP_OPTERR - Stop on Errors
@findex ERRSTOP_OPTERR

After invoking this macro, if @code{optionProcess()}
encounters an error, it will call @code{exit(1)} rather than return.
This is the default processing mode.  It can be overridden by
specifying @code{allow-errors} in the definitions file,
or invoking the macro @xref{ERRSKIP_OPTERR}.

@node HAVE_OPT
@subsection HAVE_OPT( <NAME> ) - Have this option?
@findex HAVE_OPT

This macro yields true if the option has been specified
in any fashion at all.  It is used thus:

@example
if (HAVE_OPT( NAME )) @{
    <do-things-associated-with-opt-name>;
@}
@end example

@node ISSEL_OPT
@subsection ISSEL_OPT( <NAME> ) - Is Option Selected?
@findex ISSEL_OPT

This macro yields true if the option has been
specified either on the command line or via a SET/DISABLE macro.

@node ISUNUSED_OPT
@subsection ISUNUSED_OPT( <NAME> ) - Never Specified?
@findex ISUNUSED_OPT

This macro yields true if the option has
never been specified, or has been cleared via the
@code{CLEAR_OPT()} macro.

@node OPTION_CT
@subsection OPTION_CT - Full Count of Options
@findex OPTION_CT

The full count of all options, both those defined
and those generated automatically by AutoOpts.  This is primarily
used to initialize the program option descriptor structure.

@node OPT_ARG
@subsection OPT_ARG( <NAME> ) - Option Argument String
@findex OPT_ARG

The option argument value as a pointer to string.  Note that argument
values that have been specified as numbers are stored as numbers or
keywords.  For such options, use instead the @code{OPT_VALUE_name}
define.  It is used thus:

@example
if (HAVE_OPT( NAME )) @{
    char* p = OPT_ARG( NAME );
    <do-things-with-opt-name-argument-string>;
@}
@end example

@node OPT_NO_XLAT_CFG_NAMES
@subsection OPT_NO_XLAT_CFG_NAMES - option name xlation
@findex OPT_NO_XLAT_CFG_NAMES

Invoking this macro will disable the translation of option names only while
processing configuration files and environment variables.  This must be
invoked before the first call to @code{optionProcess}..  You need not invoke
this if your option definition file contains the attribute assignment,
``@code{no-xlate = opt-cfg;}''.

@node OPT_NO_XLAT_OPT_NAMES
@subsection OPT_NO_XLAT_OPT_NAMES - option name xlation
@findex OPT_NO_XLAT_OPT_NAMES

Invoking this macro will completely disable the translation of option names.
This must be invoked before the first call to @code{optionProcess}.  You need
not invoke this if your option definition file contains the attribute
assignment, ``@code{no-xlate = opt;}''.

@node OPT_VALUE_name
@subsection OPT_VALUE_name - Option Argument Value
@findex OPT_VALUE_name

This macro gets emitted only for options that take numeric, keyword or set
membership arguments.  The macro yields a word-sized integer containing the
enumeration, bit set or numeric value for the option argument.

@example
int opt_val = OPT_VALUE_name;
@end example

@node OPT_XLAT_CFG_NAMES
@subsection OPT_XLAT_CFG_NAMES - option name xlation
@findex OPT_XLAT_CFG_NAMES

If @code{ENABLE_NLS} is defined and @code{no-xlate} has been not set to the
value @emph{anything}, this macro will cause the translation of option names
to happen before starting the processing of configuration files and
environment variables.  This will change the recognition of options within the
@code{$PROGRAMNAME} environment variable, but will not alter the names used
for setting options via @code{$PROGRAMNAME_name} environment variables.

This must be invoked before the first call to @code{optionProcess}.  You might
need to use this macro if your option definition file contains the attribute
assignment, ``@code{no-xlate = opt;}'' or ``@code{no-xlate = opt-cfg;}'', and
you have determined in some way that you wish to override that.

@node OPT_XLAT_OPT_NAMES
@subsection OPT_XLAT_OPT_NAMES - option name xlation
@findex OPT_XLAT_OPT_NAMES

If @code{ENABLE_NLS} is defined and @code{no-xlate} has been not set to the
value @emph{anything}, translate the option names before processing the
command line options.  Long option names may thus be localized.  (If the names
were translated before configuration processing, they will not be
re-translated.)

This must be invoked before the first call to @code{optionProcess}.  You might
need to use this macro if your option definition file contains the attribute
assignment, ``@code{no-xlate = opt;}'' and you have determined in some way that
you wish to override that.

@node RESTART_OPT
@subsection RESTART_OPT( n ) - Resume Option Processing
@findex RESTART_OPT

If option processing has stopped (either because of an error
or something was encountered that looked like a program argument),
it can be resumed by providing this macro with the index @code{n}
of the next option to process and calling @code{optionProcess()} again.

@node SET_OPT_name
@subsection SET_OPT_name - Force an option to be set
@findex SET_OPT_name

This macro gets emitted only when the given
option has the @code{settable} attribute specified.

The form of the macro will actually depend on whether the option is
equivalenced to another, has an option argument and/or has an assigned
handler procedure.  If the option has an argument, then this macro will
too.  Beware that the argument is not reallocated, so the value must not
be on the stack or deallocated in any other way for as long as the value
might get referenced.

If you have supplied at least one @file{homerc} file
(@pxref{program attributes}), this macro will be emitted for the
@code{--save-opts} option.

@example
SET_OPT_SAVE_OPTS( "filename" );
@end example

@noindent
@xref{automatic options}, for a discussion of the implications of using
this particular example.

@node STACKCT_OPT
@subsection STACKCT_OPT( <NAME> ) - Stacked Arg Count
@findex STACKCT_OPT

When the option handling attribute is specified
as @code{stack_arg}, this macro may be used to determine how
many of them actually got stacked.

Do not use this on options that have not been stacked or has not been
specified (the @code{stack_arg} attribute must have been specified,
and @code{HAVE_OPT(<NAME>)} must yield TRUE).
Otherwise, you will likely seg fault.

@example
if (HAVE_OPT( NAME )) @{
    int     ct = STACKCT_OPT(  NAME );
    char**  pp = STACKLST_OPT( NAME );

    do  @{
        char* p = *pp++;
        do-things-with-p;
    @} while (--ct > 0);
@}
@end example

@node STACKLST_OPT
@subsection STACKLST_OPT( <NAME> ) - Argument Stack
@findex STACKLST_OPT

The address of the list of pointers to the
option arguments.  The pointers are ordered by the order in
which they were encountered in the option presets and
command line processing.

Do not use this on options that have not been stacked or has not been
specified (the @code{stack_arg} attribute must have been specified,
and @code{HAVE_OPT(<OPTION>)} must yield TRUE).
Otherwise, you will likely seg fault.

@example
if (HAVE_OPT( NAME )) @{
    int     ct = STACKCT_OPT(  NAME );
    char**  pp = STACKLST_OPT( NAME );

    do  @{
        char* p = *pp++;
        do-things-with-p;
    @} while (--ct > 0);
@}
@end example

@node START_OPT
@subsection START_OPT - Restart Option Processing
@findex START_OPT

This is just a shortcut for RESTART_OPT(1) (@xref{RESTART_OPT}.)

@node STATE_OPT
@subsection STATE_OPT( <NAME> ) - Option State
@findex STATE_OPT

If you need to know if an option was set because of presetting actions
(configuration file processing or environment variables), versus a command
line entry versus one of the SET/DISABLE macros, then use this macro.  It
will yield one of four values: @code{OPTST_INIT}, @code{OPTST_SET},
@code{OPTST_PRESET} or @code{OPTST_DEFINED}.  It is used thus:

@example
switch (STATE_OPT( NAME )) @{
    case OPTST_INIT:
        not-preset, set or on the command line.  (unless CLEAR-ed)

    case OPTST_SET:
        option set via the SET_OPT_NAME() macro.

    case OPTST_PRESET:
        option set via an configuration file or environment variable

    case OPTST_DEFINED:
        option set via a command line option.

    default:
        cannot happen :)
@}
@end example

@node USAGE
@subsection USAGE( exit-code ) - Usage invocation macro
@findex USAGE

This macro invokes the procedure registered to display
the usage text.  Normally, this will be @code{optionUsage} from the
AutoOpts library, but you may select another procedure by specifying
@code{usage = "proc_name"} program attribute.  This procedure must
take two arguments@:  first, a pointer to the option descriptor, and
second the exit code.  The macro supplies the option descriptor
automatically.  This routine is expected to call @code{exit(3)} with
the provided exit code.

The @code{optionUsage} routine also behaves differently depending
on the exit code:

@table @code
@item EXIT_SUCCESS (the value zero)
It is assumed that full usage help has been requested.  Consequently, more
information is provided than when displaying usage and exiting with a
non-zero exit code.  Output will be sent to @file{stdout} and the program will
exit with a zero status code.

@item EX_USAGE (64)
The abbreviated usage will be printed to @file{stdout} and the program will
exit with a zero status code.  ``EX_USAGE'' may or may not be 64.  If your
system provides ``/usr/include/sysexits.h'' that has a different value,
then that value will be used.

@item any other value
The abbreviated usage will be printed to stderr and the program will
exit with the provided status code.
@end table

@node VALUE_OPT_name
@subsection VALUE_OPT_name - Option Flag Value
@findex VALUE_OPT_name

This is a #define for the flag character used to
specify an option on the command line.  If @code{value} was not
specified for the option, then it is a unique number associated
with the option.  @code{option value} refers to this value,
@code{option argument} refers to the (optional) argument to the
option.

@example
switch (WHICH_OPT_OTHER_OPT) @{
case VALUE_OPT_NAME:
    this-option-was-really-opt-name;
case VALUE_OPT_OTHER_OPT:
    this-option-was-really-other-opt;
@}
@end example

@node VERSION
@subsection VERSION - Version and Full Version
@findex VERSION

If the @code{version} attribute is defined for the program,
then a stringified version will be #defined as PROGRAM_VERSION and
PROGRAM_FULL_VERSION.  PROGRAM_FULL_VERSION is used for printing
the program version in response to the version option.  The version
option is automatically supplied in response to this attribute, too.

You may access PROGRAM_VERSION via @code{programOptions.pzFullVersion}.

@node WHICH_IDX_name
@subsection WHICH_IDX_name - Which Equivalenced Index
@findex WHICH_IDX_name

This macro gets emitted only for equivalenced-to options.  It is used to
obtain the index for the one of the several equivalence class members
set the equivalenced-to option.

@example
switch (WHICH_IDX_OTHER_OPT) @{
case INDEX_OPT_NAME:
    this-option-was-really-opt-name;
case INDEX_OPT_OTHER_OPT:
    this-option-was-really-other-opt;
@}
@end example

@node WHICH_OPT_name
@subsection WHICH_OPT_name - Which Equivalenced Option
@findex WHICH_OPT_name

This macro gets emitted only for equivalenced-to options.  It is used to
obtain the value code for the one of the several equivalence class members
set the equivalenced-to option.

@example
switch (WHICH_OPT_OTHER_OPT) @{
case VALUE_OPT_NAME:
    this-option-was-really-opt-name;
case VALUE_OPT_OTHER_OPT:
    this-option-was-really-other-opt;
@}
@end example

@node teOptIndex
@subsection teOptIndex - Option Index and Enumeration
@findex teOptIndex

This enum defines the complete set of options, both
user specified and automatically provided.  This can be used,
for example, to distinguish which of the equivalenced options
was actually used.

@example
switch (pOptDesc->optActualIndex) @{
case INDEX_OPT_FIRST:
    stuff;
case INDEX_OPT_DIFFERENT:
    different-stuff;
default:
    unknown-things;
@}
@end example

@node OPTIONS_STRUCT_VERSION
@subsection OPTIONS_STRUCT_VERSION - active version

You will not actually need to reference this value, but you need to be
aware that it is there.  It is the first value in the option descriptor
that you pass to @code{optionProcess}.  It contains a magic number and
version information.  Normally, you should be able to work with a more
recent option library than the one you compiled with.  However, if the
library is changed incompatibly, then the library will detect the out of
date magic marker, explain the difficulty and exit.  You will then need
to rebuild and recompile your option definitions.  This has rarely been
necessary.

@ignore
END   == AUTOOPTS-API == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == AUTOOPTS-DATA == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c === SECTION MARKER

@node Multi-Threading
@section Multi-Threading

AutoOpts was designed to configure a program for running.  This generally
happens before much real work has been started.  Consequently, it is
expected to be run before multi-threaded applications have started multiple
threads.  However, this is not always the case. Some applications may
need to reset and reload their running configuration, and some may use
@code{SET_OPT_xxx()} macros during processing.  If you need to dynamically
change your option configuration in your multi-threaded application, it is
your responsibility to prevent all threads from accessing the option
configuration state, except the one altering the configuration.

The various accessor macros (@code{HAVE_OPT()}, etc.) do not modify state
and are safe to use in a multi-threaded application.  It is safe as long
as no other thread is concurrently modifying state, of course.

@c === SECTION MARKER

@node option descriptor
@section Option Descriptor File
@cindex option descriptor

This is the module that is to be compiled and linked with your program.
It contains internal data and procedures subject to change.  Basically,
it contains a single global data structure containing all the
information provided in the option definitions, plus a number of static
strings and any callout procedures that are specified or required.  You
should never have need for looking at this, except, perhaps, to examine
the code generated for implementing the @code{flag-code} construct.

@c === SECTION MARKER

@node Using AutoOpts
@section Using AutoOpts
@cindex using AutoOpts

There are actually several levels of ``using'' autoopts.
Which you choose depends upon how you plan to distribute
(or not) your application.

@menu
* local use::               local-only use
* binary not installed::    binary distro, AutoOpts not installed
* binary pre-installed::    binary distro, AutoOpts pre-installed
* source pre-installed::    source distro, AutoOpts pre-installed
* source not installed::    source distro, AutoOpts not installed
@end menu

@node local use
@subsection local-only use

To use AutoOpts in your application where you do not have to
worry about distribution issues, your issues are simple and few.

@itemize @bullet
@item
Create a file @samp{myopts.def}, according to the documentation above.
It is probably easiest to start with the example in @ref{Quick Start}
and edit it into the form you need.

@item
Run AutoGen to create the option interface file (@code{myopts.h})
and the option descriptor code (@code{myopts.c}):

@example
autogen myopts.def
@end example

@item
In all your source files where you need to refer to option state,
@code{#include "myopts.h"}.
@item
In your main routine, code something along the lines of:

@example
#define ARGC_MIN some-lower-limit
#define ARGC_MAX some-upper-limit
main( int argc, char** argv )
@{
    @{
        int arg_ct = optionProcess( &myprogOptions, argc, argv );
        argc -= arg_ct;
        if ((argc < ARGC_MIN) || (argc > ARGC_MAX)) @{
            fprintf( stderr, "%s ERROR:  remaining args (%d) "
                     "out of range\n", myprogOptions.pzProgName,
                     argc );

            USAGE( EXIT_FAILURE );
        @}
        argv += arg_ct;
    @}
    if (HAVE_OPT(OPTN_NAME))
        respond_to_optn_name();
    ...
@}
@end example

@item
Compile @samp{myopts.c} and link your program
with the following additional arguments:

@example
`autoopts-config cflags ldflags` myopts.c
@end example
@end itemize

@node binary not installed
@subsection binary distro, AutoOpts not installed

If you will be distributing (or copying) your project to a system that
does not have AutoOpts installed, you will need to statically link the
AutoOpts library, ``libopts'' into your program.  Get the link information
with ``@code{static-libs}'' instead of ``@code{ldflags}'':

@example
`autoopts-config static-libs`
@end example

@node binary pre-installed
@subsection binary distro, AutoOpts pre-installed

If you will be distributing (or copying) your project to a system that does
have AutoOpts (or only ``libopts'') installed, you will still need to ensure
that the library is findable at program load time, or you will still have to
statically link.  The former can be accomplished by linking your project with
@code{--rpath} or by setting the @code{LD_LIBRARY_PATH} appropriately.
Otherwise, @xref{binary not installed}.

@node source pre-installed
@subsection source distro, AutoOpts pre-installed

If you will be distributing your project to a system that will build
your product but it may not be pre-installed with AutoOpts, you will
need to do some configuration checking before you start the build.
Assuming you are willing to fail the build if AutoOpts has not been
installed, you will still need to do a little work.

AutoOpts is distributed with a configuration check M4 script,
@file{autoopts.m4}.  It will add an @code{autoconf} macro named,
@code{AG_PATH_AUTOOPTS}.  Add this to your @file{configure.ac} script
and use the following substitution values:

@table @code
@item AUTOGEN
the name of the autogen executable
@item AUTOGEN_TPLIB
the directory where AutoGen template library is stored
@item AUTOOPTS_CFLAGS
the compile time options needed to find the AutoOpts headers
@item AUTOOPTS_LIBS
the link options required to access the @code{libopts} library
@end table

@node source not installed
@subsection source distro, AutoOpts not installed

If you will be distributing your project to a system that will build
your product but it may not be pre-installed with AutoOpts, you may
wish to incorporate the sources for @code{libopts} in your project.
To do this, I recommend reading the tear-off libopts library
@file{README} that you can find in the @file{pkg/libopts} directory.
You can also examine an example package (blocksort) that incorporates
this tear off library in the autogen distribution directory.  There is
also a web page that describes what you need to do:
@example
@url{http://autogen.sourceforge.net/blocksort.html}
@end example

Alternatively, you can pull the @code{libopts} library sources into
a build directory and build it for installation along with your package.
This can be done approximately as follows:
@example
tar -xzvf `autoopts-config libsrc`
cd libopts-*
./bootstrap
configure
make
make install
@end example
That will install the library, but not the headers or anything else.

@c === SECTION MARKER

@node Presetting Options
@section Configuring your program
@cindex shell options

AutoOpts supports the notion of ``presetting'' the value or state of an option.
The values may be obtained either from environment variables or from
configuration files (@file{rc} or @file{ini} files).  In order to take
advantage of this, the AutoOpts client program must specify these features
in the option descriptor file (@pxref{program attributes}) with the
@code{rcfile} or @code{environrc} attributes.

@menu
* loading rcfile::      configuration file presets
* saving rcfile::       Saving the presets into a configuration file
* sample rcfile::       Creating a sample configuration file
* environrc::           environment variable presets
* config example::      Config file only example
@end menu

It is also possible to configure your program @i{without} using
the command line option parsing code.  This is done by using
only the following four functions from the @file{libopts} library:

@table @samp
@item configFileLoad
(@pxref{libopts-configFileLoad}) will parse the contents of a config
file and return a pointer to a structure representing the hierarchical
value.  The values are sorted alphabetically by the value name and all
entries with the same name will retain their original order.
Insertion sort is used.

@item optionGetValue
(@pxref{libopts-optionGetValue}) will find the first value within the
hierarchy with a name that matches the name passed in.

@item optionNextValue
(@pxref{libopts-optionNextValue}) will return the next value that
follows the value passed in as an argument.  If you wish to get all
the values for a particular name, you must take note when the name
changes.

@item optionUnloadNested
(@pxref{libopts-optionUnloadNested}).  The pointer passed in must be
of type, @code{OPARG_TYPE_HIERARCHY} (see the autoopts/options.h
header file).  @code{configFileLoad} will return a @code{tOptionValue}
pointer of that type.  This function will release all the associated
memory.  @code{AutoOpts} generated code uses this function for its own
needs.  Client code should only call this function with pointers
gotten from @code{configFileLoad}.
@end table

@node loading rcfile
@subsection configuration file presets
@cindex rcfile

Configuration files are enabled by specifying the program attribute
@code{homerc} (@pxref{program attributes}).  Any option not marked
with the ``no-preset'' attribute may appear in a configuration file.
The files loaded are selected both by the @code{homerc} entries and,
optionally, via a command line option.  The first component of the
@code{homerc} entry may be an environment variable such as @code{$HOME}, or
it may also be @code{$$} (@strong{two} dollar sign characters) to specify
the directory of the executable.  For example:

@example
homerc = "$$/../share/autogen";
@end example

@noindent
will cause the AutoOpts library to look in the normal autogen datadir
relative to the current installation directory for autogen.

The configuration files are processed in the order they are specified by
the @code{homerc} attribute, so that each new file will normally override
the settings of the previous files.  This may be overridden by marking some
options for @code{immediate action} (@pxref{Immediate Action}).  Any such
options are acted upon in @strong{reverse} order.  The disabled
@code{load-opts} (@code{--no-load-opts}) option, for example, is an
immediate action option.  Its presence in the last @code{homerc} file will
prevent the processing of any prior @code{homerc} files because its effect
is immediate.

Configuration file processing can be completely suppressed by specifying
@code{--no-load-opts} on the command line, or @code{PROGRAM_LOAD_OPTS=no} in
the environment (if @code{environrc} has been specified).

See the ``Configuration File Format'' section (@pxref{Config File Format})
for details on the format of the file.

@node saving rcfile
@subsection Saving the presets into a configuration file

When configuration files are enabled for an application, the user is
also provided with an automatically supplied @code{--save-opts} option.
All of the known option state will be written to either the specified
output file or, if it is not specified, then to the last specified
@code{homerc} file.

@node sample rcfile
@subsection Creating a sample configuration file
@cindex sample rcfile

AutoOpts is shipped with a template named, @file{rc-sample.tpl}.
If your option definition file specifies the @code{homerc} attribute,
then you may invoke @file{autogen} thus:

@example
autogen -Trc-sample <your-option-def-file>
@end example

This will, by default, produce a sample file named,
@file{sample-<prog-name>rc}.  It will be named differently if you specify your
configuration (rc) file name with the @code{rcfile} attribute.  In that case,
the output file will be named, @file{sample-<rcfile-name>}.  It will contain
all of the program options not marked as @code{no-preset}.  It will also
include the text from the @code{doc} attribute.

@ignore
END   == AUTOOPTS-DATA == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == AO-DATA1 == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@node environrc
@subsection environment variable presets
@cindex environrc

If the AutoOpts client program specifies @code{environrc} in its
option descriptor file, then environment variables will be used for
presetting option state.  Variables will be looked for that are named,
@code{PROGRAM_OPTNAME} and @code{PROGRAM}.  @code{PROGRAM} is the
upper cased @code{C-name} of the program, and @code{OPTNAME} is the
upper cased @code{C-name} of a specific option.  (The @code{C-name}s
are the regular names with all special characters converted to
underscores (@code{_}).)

Option specific environment variables are processed after (and thus
take precedence over) the contents of the @code{PROGRAM} environment
variable.  The option argument string for these options takes on the
string value gotten from the environment.  Consequently, you can only
have one instance of the @code{OPTNAME}.

If a particular option may be disabled, then its disabled state is
indicated by setting the @code{PROGRAM_OPTNAME} value to the
disablement prefix.  So, for example, if the disablement prefix were
@code{dont}, then you can disable the @code{optname} option by setting
the @code{PROGRAM_OPTNAME}' environment variable to `@i{dont}'.
@xref{Common Attributes}.

The @code{PROGRAM} environment string is tokenized and parsed much
like a command line.  Doubly quoted strings have backslash escapes
processed the same way they are processed in C program constant
strings.  Singly quoted strings are ``pretty raw'' in that backslashes are
honored before other backslashes, apostrophes, newlines and cr/newline
pairs.  The options must be introduced with hyphens in the same way as
the command line.

Note that not all options may be preset.  Options that are specified with the
@code{no-preset} attribute and the @code{--help}, @code{--more-help},
and @code{--save-opts} auto-supported options may not be preset.

@node config example
@subsection Config file only example
@cindex rcfile
@cindex Configuration File
@cindex Configuration File example

If for some reason it is difficult or unworkable to integrate configuration
file processing with command line option parsing, the @code{libopts}
(@pxref{libopts procedures}) library can still be used to process configuration
files.  Below is a ``@t{Hello, World!}'' greeting program that tries
to load a configuration file @file{hello.conf} to see if it should use
an alternate greeting or to personalize the salutation.
@ignore
END   == AO-DATA1 == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == AO-DATA2 == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c === SECTION MARKER

@node Config File Format
@section Configuration File Format
@cindex Configuration File

The configuration file is designed to associate names and values, much like
an AutoGen Definition File (@pxref{Definitions File}).  Unfortunately, the
file formats are different.  Specifically, AutoGen Definitions provide for
simpler methods for the precise control of a value string and provides for
dynamically computed content.  Configuration files have some established
traditions in their layout.  So, they are different, even though they do
both allow for a single name to be associated with multiple values and they
both allow for hierarchical values.

@menu
* config name/string-value::    assigning a string value to a configurable
* config integer-values::       integer values
* config nested-values::        hierarchical values
* config directives::           configuration file directives
* config comments::             comments in the configuration file
@end menu

@node config name/string-value
@subsection assigning a string value to a configurable

The basic syntax is a name followed by a value on a single line.  They are
separated from each other by either white space, a colon (@code{:}) or an
equal sign (@code{=}).  The colon or equal sign may optionally be surrounded
by additional white space.  If more than one value line is needed, a
backslash (@code{\}) may be used to continue the value.  The backslash (but
not the newline) will be erased.  Leading and trailing white space is always
stripped from the value.

Fundamentally, it looks like this:

@example
name  value for that name
name = another \
     multi-line value \
     for that name.
name: a *third* value for ``name''
@end example

If you need more control over the content of the value, you may enclose the
value in XML style brackets:
@example
<name>value </name>
@end example
@noindent
Within these brackets you need not (must not) continue the value data with
backslashes.  You may also select the string formation rules to use, just
add the attribute after the name, thus: @code{<name keep>}.

@table @samp
@item keep
This mode will keep all text between the brackets and not strip any
white space.
@item uncooked
This mode strips leading and trailing white space, but not do any
quote processing.  This is the default and need not be specified.
@item cooked
The text is trimmed of leading and trailing white space and XML encodings
are processed.  These encodings are slightly expanded over the XML
specification.  They are specified with an ampersand followed by a value
name or numeric value and then a semicolon:

@table @samp
@item  amp
@itemx lt
@itemx gt
@itemx quot
@itemx apos
@itemx #dd
@itemx #xHH

These are all per fairly standad HTML and/or XML encodings.
Additionally:

@item bs
The ASCII back space character.
@item ff
The ASCII form feed character.
@item ht
The ASCII horizontal (normal) tab character.
@item cr
The ASCII carriage return character.
@item vt
The ASCII vertical tab character.
@item bel
The ASCII alarm bell character.
@item nl
The ASCII new line character.
@item space
The ASCII space character.  Normally not necessary, but if you want
to preserve leading or trailing space characters, then use this.
@end table
@end table

And here is an example of an XML-styled value:

@example
<name cooked>
    This is&nl;&ht;another multi-line
&ht;string example.
</name>
@end example

The string value associated with ``name'' will be exactly the text enclosed
in quotes with the encoded characters ``cooked'' as you would expect
(three text lines with the last line not ending with a newline, but
ending with a period).

@node config integer-values
@subsection integer values

A name can be specified as having an integer value.  To do this, you
must use the XML-ish format and specify a ``type'' attribute for
the name:

@example
<name type=integer> 1234 </name>
@end example

Boolean, enumeration and set membership types will be added as time
allows.  ``type=string'' is also supported, but also is the default.

@node config nested-values
@subsection hierarchical values

In order to specify a hierarchical value, you *must* use XML-styled
formatting, specifying a type that is shorter and easier to spell:

@example
<structured-name type=nested>
    [[....]]
</structured-name>
@end example

@noindent
The ellipsis may be filled with any legal configuration file name/value
assignments.

@node config directives
@subsection configuration file directives
@cindex autoopts directives

The @code{<?} marker indicates an XML directive.
There is only one directive supported:  program sectioning,
though two syntaxes are supported.

If, for example, you have a collection of programs that work closely
together and, likely, have a common set of options, these programs may use a
single, sectioned, configuration file.  The file may be sectioned in either
of two ways.  The two ways may not be intermixed in a single configuration
file.  All text before the first segmentation line is processed, then only
the segment that applies:

@table @samp
@item <?auto-options ...>
The @code{...} ellipsis may contain AutoOpts option processing options.
Currently, that consists of one or both of:

@table @code
@item gnu
@itemx autoopts
to indicate GNU-standard or AutoOpts-standard layout of usage and
version information, and/or

@item misuse-usage
@itemx no-misuse-usage
to indicate whether the available options should be listed when
an invalid option appears on the command line.
@end table
@noindent
Anything else will be silently ignored.

@item <?program prog-name>
The @code{<?} marker indicates an XML directive.
The file is partitioned by these lines and the options are processed
for the @code{prog-name} program only before the first @code{<?program}
directive and the program section with a matching program name.

@item [PROG_NAME]
This is basically an alias for @code{<?program prog-name>}, except that
the program name must be upper cased and segmented only with underscores.
@end table

@noindent
Segmentation does not apply if the config file is being parsed with
the @code{configFileLoad(3AutoOpts)} function.

@node config comments
@subsection comments in the configuration file

Comments are lines beginning with a hash mark (@code{#}),
XML-style comments (@code{<!-- arbitrary text -->}), and
unrecognized XML directives.

@example
# this is a comment
<!-- this is also
     a comment -->
<?this is
  a bad comment ;->
@end example

@c === SECTION MARKER

@node shell options
@section AutoOpts for Shell Scripts
@cindex shell options
@cindex configuration file

AutoOpts may be used with shell scripts either by automatically creating a
complete program that will process command line options and pass back
the results to the invoking shell by issuing shell variable assignment
commands, or it may be used to generate portable shell code that can
be inserted into your script.

The functionality of these features, of course, is somewhat constrained
compared with the normal program facilities.  Specifically, you cannot
invoke callout procedures with either of these methods.  Additionally,
if you generate a shell script to do the parsing:

@enumerate
@item
You cannot obtain options from configuration files.
@item
You cannot obtain options from environment variables.
@item
You cannot save the option state to an option file.
@item
Option conflict/requirement verification is disabled.
@end enumerate

Both of these methods are enabled by running AutoGen on
the definitions file with the additional main procedure attribute:

@example
main = @{ main-type = shell-process; @};
@end example
@noindent
or:
@example
main = @{ main-type = shell-parser; @};
@end example

If you do not supply a @code{proc-to-call}, it will default to
@code{optionPutShell}.  That will produce a program that will process the
options and generate shell text for the invoking shell to interpret
(@pxref{binary-parser}).  If you supply the name, @code{optionParseShell}, then
you will have a program that will generate a shell script that can parse the
options (@pxref{script-parser}).  If you supply a different procedure name,
you will have to provide that routine and it may do whatever you like.

@menu
* binary-parser::        Parsing with an Executable
* script-parser::        Parsing with a Portable Script
@end menu

@node binary-parser
@subsection Parsing with an Executable

The following commands are approximately all that is needed
to build a shell script command line option parser from
an option definition file:

@example
autogen -L <opt-template-dir> test-errors.def
cc -o test-errors -L <opt-lib-dir> -I <opt-include-dir> \
        -DTEST_PROGRAM_OPTS test-errors.c -lopts
@end example

The resulting program can then be used within your shell script as follows:

@example
eval `./test-errors "$@@"`
if [ -z "$@{OPTION_CT@}" ] ; then exit 1 ; fi
test $@{OPTION_CT@} -gt 0 && shift $@{OPTION_CT@}
@end example
@ignore
END   == AO-DATA2 == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@ignore
START == AUTOINFO == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c === SECTION MARKER

@node AutoInfo
@section Automated Info Docs
@cindex AutoInfo

AutoOpts provides two templates for producing @file{.texi} documentation.
@file{agtexi-cmd.tpl} for the invoking section, and @file{aginfo3.tpl} for
describing exported library functions and macros.

For both types of documents, the documentation level is selected by
passing a @samp{-DLEVEL=<level-name>} argument to AutoGen when you build
the document.  (See the example invocation below.)

Two files will be produced, a @file{.texi} file and a @file{.menu} file.
You should include the text in the @file{.menu} file in a @file{@@menu}
list, either with @file{@@include}-ing it or just copying text.
The @file{.texi} file should be @file{@@include}-ed where the invoking
section belongs in your document.

The @file{.texi} file will contain an introductory paragraph, a menu
and a subordinate section for the invocation usage and for each
documented option.  The introductory paragraph is normally the boiler
plate text, along the lines of:

@example
This chapter documents the @@file@{AutoOpts@} generated usage text
and option meanings for the @@file@{your-program@} program.
@end example

@noindent
or:

@example
These are the publicly exported procedures from the lib@i{name} library.
Any other functions mentioned in the @i{header} file are for the private use
of the library.
@end example

@menu
* command-info::      ``invoking'' info docs
* library-info::      library info docs
@end menu

@node command-info
@subsection ``invoking'' info docs

Using the option definitions for an AutoOpt client program, the
@file{agtexi-cmd.tpl} template will produce texinfo text that documents the
invocation of your program.  The text emitted is designed to be included
in the full texinfo document for your product.  It is not a stand-alone
document.  The usage text for the @ref{autogen usage},
@ref{getdefs usage} and @ref{columns usage} programs, are included in
this document and are all generated using this template.

If your program's option definitions include a
@samp{prog-info-descrip} section, then that text will replace the
boilerplate introductory paragraph.

@noindent
These files are produced by invoking the following command:

@example
autogen -L $@{prefix@}/share/autogen -Tagtexi-cmd.tpl \
        -DLEVEL=section your-opts.def
@end example

@noindent
Where @file{$@{prefix@}} is the AutoGen installation prefix
and @file{your-opts.def} is the name of your product's option
definition file.

@node library-info
@subsection library info docs

The @file{texinfo} doc for libraries is derived from mostly the same
information as is used for producing man pages @xref{man3}.  The main
difference is that there is only one output file and the individual
functions are referenced from a @code{.texi} menu.  There is also
a small difference in the global attributes used:

@multitable @columnfractions .02 .23 .65
@item @tab lib_description
@tab A description of the library.  This text appears before the menu.
If not provided, the standard boilerplate version will be inserted.
@item
@item @tab see_also
@tab The @code{SEE ALSO} functionality is not supported for the
@file{texinfo} documentation, so any @code{see_also} attribute will be ignored.
@end multitable

@noindent
These files are produced by invoking the following commands:

@example
getdefs linenum srcfile template=aginfo3.tpl output=libexport.def \
       <source-file-list>

autogen -L $@{prefix@}/share/autogen -DLEVEL=section libexport.def
@end example

@noindent
Where @file{$@{prefix@}} is the AutoGen installation prefix
and @file{libexport.def} is some name that suits you.

An example of this can be seen in this document, @xref{libopts procedures}.

@c === SECTION MARKER

@node AutoMan pages
@section Automated Man Pages
@cindex AutoMan pages

AutoOpts provides two templates for producing man pages.
The command (@file{man1}) pages are derived from the options definition
file, and the library (@file{man3}) pages are derived from
stylized comments (@pxref{getdefs Invocation}).

@menu
* man1::      command line man pages
* man3::      library man pages
@end menu

@node man1
@subsection command line man pages

Using the option definitions for an AutoOpts client program,
the @samp{agman-cmd.tpl} template will produce an nroff document
suitable for use as a @samp{man(1)} page document for a command
line command.  The description section of the document is either
the @samp{prog-man-descrip} text, if present, or the @samp{detail}
text.

Each option in the option definitions file is fully documented
in its usage.  This includes all the information documented
above for each option (@pxref{option attributes}), plus
the @samp{doc} attribute is appended.  Since the @samp{doc}
text is presumed to be designed for @code{texinfo} documentation,
@code{sed} is used to convert some constructs from @code{texi}
to @code{nroff}-for-@code{man}-pages.  Specifically,

@example
convert @@code, @@var and @@samp into \fB...\fP phrases
convert @@file into \fI...\fP phrases
Remove the '@@' prefix from curly braces
Indent example regions
Delete the example commands
Replace @samp{end example} command with ".br"
Replace the @samp{@@*} command with ".br"
@end example

@noindent
This document is produced by invoking the following command:

@example
autogen -L $@{prefix@}/share/autogen -Tagman-cmd.tpl options.def
@end example

@noindent
Where @file{$@{prefix@}} is the AutoGen installation prefix and
@file{options.def} is the name of your product's option definition file.
I do not use this very much, so any feedback or improvements would be
greatly appreciated.

@node man3
@subsection library man pages

Two global definitions are required, and then
one library man page is produced for each @code{export_func} definition
that is found.  It is generally convenient to place these definitions
as @file{getdefs} comments (@pxref{getdefs Invocation}) near the procedure
definition, but they may also be a separate AutoGen definitions file
(@pxref{Definitions File}).  Each function will be cross referenced
with their sister functions in a @file{SEE ALSO} section.  A global
@code{see_also} definition will be appended to this cross referencing text.

@noindent
The two global definitions required are:

@multitable @columnfractions .02 .15 .77
@item @tab library
@tab This is the name of your library, without the @file{lib} prefix.
The AutoOpts library is named @file{libopts.so...}, so the @code{library}
attribute would have the value @code{opts}.
@item
@item @tab header
@tab Generally, using a library with a compiled program entails
@code{#include}-ing a header file.  Name that header with this attribute.
In the case of AutoOpts, it is generated and will vary based on the
name of the option definition file.  Consequently, @file{your-opts.h} is
specified.
@end multitable

@noindent
The @code{export_func} definition should contain the following attributes:

@multitable @columnfractions .02 .15 .77
@item @tab name
@tab The name of the procedure the library user may call.
@item @tab what
@tab A brief sentence describing what the procedure does.
@item @tab doc
@tab A detailed description of what the procedure does.
It may ramble on for as long as necessary to properly describe it.
@item @tab err
@tab A short description of how errors are handled.
@item @tab ret_type
@tab The data type returned by the procedure.
Omit this for @code{void} procedures.
@item @tab ret_desc
@tab Describe what the returned value is, if needed.
@item @tab private
@tab If specified, the function will @strong{not} be documented.
This is used, for example, to produce external declarations for functions
that are not available for public use, but are used in the generated text.
@item
@item @tab arg
@tab This is a compound attribute that contains:
@end multitable
@multitable @columnfractions .02 .15 .15 .62
@item @tab @tab arg_type
@tab The data type of the argument.
@item @tab @tab arg_name
@tab A short name for it.
@item @tab @tab arg_desc
@tab A brief description.
@end multitable

@noindent
As a @file{getdefs} comment, this would appear something like this:

@example
/*=--subblock=arg=arg_type,arg_name,arg_desc =*/
/*=*
 * library: opts
 * header:  your-opts.h
=*/
/*=export_func optionProcess
 *
 * what: this is the main option processing routine
 * arg:  + tOptions* + pOpts + program options descriptor +
 * arg:  + int       + argc  + program arg count  +
 * arg:  + char**    + argv  + program arg vector +
 * ret_type:  int
 * ret_desc:  the count of the arguments processed
 *
 * doc:  This is what it does.
 * err:  When it can't, it does this.
=*/
@end example

@noindent
Note the @code{subblock} and @code{library} comments.
@code{subblock} is an embedded @file{getdefs}
option (@pxref{getdefs subblock}) that tells it how to parse the
@code{arg} attribute.  The @code{library} and @code{header} entries
are global definitions that apply to all the documented functions.

@c === SECTION MARKER

@node getopt_long
@section Using getopt(3C)
@cindex getopt_long

There is a template named, @code{getopt.tpl} that is distributed with
AutoOpts.  Using that template instead of @code{options.tpl} will produce
completely independent source code that will parse command line options.  It
will utilize either the standard @code{getopt(3C)} or the GNU
@code{getopt_long(3GNU)} function to drive the parsing.  Which is used is
selected by the presence or absence of the @code{long-opts} program attribute.
It will save you from being dependent upon the @code{libopts} library @i{and}
it produces code ready for internationalization.  However, it also carries
with it some limitations on the use of AutoOpts features and some requirements
on the build environment.


@menu
* getopt limitations::  getopt feature limitations
* getopt building::     getopt build requirements
@end menu

@node getopt limitations
@subsection getopt feature limitations

This list of limitations is relative to the full list of AutoOpts
supported features, @xref{Features}.

@enumerate
@item
You cannot automatically take advantage of environment variable options or
automated parsing of configuration files (``rc'' or ``ini'' files).
Consequently, the resulting code does not support @file{--load-opts} or
@file{--save-opts} options automatically.

@item
You cannot use set membership, enumerated, range checked or stacked
argument type options.  In fact, you cannot use anything that depends
upon the @code{libopts} library.  You are constrained to options that
take ``@code{string}'' arguments, though you may handle the option
argument with a callback procedure.

@item
Special disablement and/or enablement prefixes are not recognized.

@item
Generated @code{main()} procedures will not work.

@item
Option coordination with external libraries will not work.

@item
Every option must be ``settable'' because the emitted code
depends upon the @code{SET_OPT_XXX} macros having been defined.
Specify this as a global (program) attribute.

@item
You must specify a main procedure of type ``main''.  The
@file{getopt.tpl} template depends upon being able to compile the
traditional .c file into a program and get it to emit the usage text.

@item
For the same reason, the traditional option parsing table code must be
emitted @b{before} the @file{getopt.tpl} template gets expanded.

@item
The usage text is, therefore, statically defined.
@end enumerate

@node getopt building
@subsection getopt build requirements

You must supply some compile and link options via environment variables.

@table @samp
@item srcdir
In case the option definition file lives in a different directory.
@item CFLAGS
Any special flags required to compile.  The flags from
@code{autoopts-config cflags} will be included automatically.  Since
the creation of the option parsing code includes creating a program
that prints out help text, if it is necessary to include files from
various directories to compile that program, you will need to specify
those directories with ``-Idirpath'' text in the @code{CFLAGS}.
Some experimentation may be necessary in that case.

@strong{NOTE}: the ``-Idirpath'' text is only needed if your option callback
functions include code that require additional ``#include'' directives.
@item LDFLAGS
Any special flags required to link.  The flags from
@code{autoopts-config ldflags} will be included automatically.  This
is required only if additional link flags for the help text emission
program might be needed.
@item CC
This is needed only if ``@code{cc}'' cannot be found in @code{$PATH}
(or it is not the one you want).
@end table

To use this, set the exported environment variables and specify ``getopt'' as
the default template in your option definitions file (@pxref{Identification}).
You will have @i{four} new files.  Assuming your definitions were in a file
named @file{myprog-opts.def} and your program name was specified as
@file{progname}, the resulting files would be created: @file{myprog-opts.h},
@file{myprog-opts.c}, @file{getopt-progname.h} and @file{getopt-progname.c}.
You must compile and link both @file{.c} files into your program.  If there
are link failures, then you are using AutoOpts features that require the
@file{libopts} library.  You must remove these features,
@xref{getopt limitations}.

These generated files depend upon configure defines to work correctly.
Therefore, you must specify a @code{config-header} attribute
(@pxref{programming attributes}) and ensure it has @code{#defines} for
either @code{HAVE_STDINT_H} or @code{HAVE_INTTYPES_H}; either
@code{HAVE_SYS_LIMITS_H} or @code{HAVE_LIMITS_H}; and
@code{HAVE_SYSEXITS_H}, if the @file{sysexits.h} header is available.
The required header files for these defines are, respectively,
the @file{/usr/include} files named:
@itemize @bullet
@item stdint.h
@item inttypes.h
@item sys/limits.h
@item limits.h
@item sysexits.h
@end itemize

@noindent
The following header files must also exist on the build platform:
@itemize @bullet
@item sys/types.h
@item stdio.h
@item string.h
@item unistd.h -- or, for getopt_long:
@item getopt.h
@end itemize
@c === SECTION MARKER

@node i18n
@section Internationalizing AutoOpts
@cindex Internationalizing AutoOpts

The generated code for AutoOpts will enable and disable the translation of
AutoOpts run time messages.  If @code{ENABLE_NLS} is defined at compile time
and @code{no-xlate} has been not set to the value @emph{anything}, then the
@code{_()} macro may be used to specify a translation function.  If undefined,
it will default to @code{gettext(3GNU)}.  This define will also enable a
callback function that @code{optionProcess} invokes at the beginning of option
processing.  The AutoOpts @code{libopts} library will always check for this
@emph{compiled with NLS} flag, so @code{libopts} does not need to be specially
compiled.  The strings returned by the translation function will be
@code{strdup(3)-ed} and kept.  They will not be re-translated, even if the
locale changes, but they will also not be dependent upon reused or unmappable
memory.

To internationalize option processing, you should first internationalize your
program.  Then, the option processing strings can be added to your translation
text by processing the AutoOpts-generated @file{my-opts.c} file and adding the
distributed @file{po/usage-txt.pot} file.  (Also by extracting the strings
yourself from the @file{usage-txt.h} file.)  When you call
@code{optionProcess}, all of the user visible AutoOpts strings will be passed
through the localization procedure established with the @code{_()}
preprocessing macro.

All of this is @emph{dis}-abled if you specify the global attribute
@code{no-xlate} to @emph{anything}.

@c === SECTION MARKER

@node Naming Conflicts
@section Naming Conflicts
@cindex Naming Conflicts

AutoOpts generates a header file that contains many C preprocessing macros and
several external names.  For the most part, they begin with either @code{opt_}
or @code{option}, or else they end with @code{_opt}.  If this happens to
conflict with other macros you are using, or if you are compiling multiple
option sets in the same compilation unit, the conflicts can be avoided.  You
may specify an external name @code{prefix} (@pxref{program attributes}) for
all of the names generated for each set of option definitions.

Among these macros, several take an option name as a macro argument.
Sometimes, this will inconveniently conflict.  For example, if you specify an
option named, @code{debug}, the emitted code will presume that @code{DEBUG} is
not a preprocessing name.  Or also, if you are building on a Windows platform,
you may find that MicroSoft has usurped a number of user space names in its
header files.  Consequently, you will get a preprocessing error if you use,
for example, @code{HAVE_OPT(DEBUG)} or @code{HAVE_OPT(INTERNAL)}
(@pxref{HAVE_OPT}) in your code.  You may trigger an obvious warning for such
conflicts by specifying the @code{guard-option-names} attribute
(@pxref{program attributes}).  That emitted code will also @code{#undef}-ine
the conflicting name.

@node All Attribute Names
@section All Attribute Names

This is the list of all the option attributes used in the various
option processing templates.  There are several flavors of attributes,
and these are not distinguished here.

@itemize @bullet
@item
Valid, current attributes that you are encouraged to use.
@item
Internally generated attributes that you cannot use at all.
I need to prefix these with a distinguished prefix.  e.g.  ``ao-''
@item
Valid attributes, but are deprecated.  Alternates should be documented.
@end itemize

This list is derived by running many example option definitions through the
option generation and man page templates and noting which attributes are
actually used.  There may be a few that are used but not exercised in my
testing.  If so, I need to ferret those out and test them, too.

@example
aliases          allow-errors  arg-default
arg-optional     arg-range     arg-type
argument         call-proc     code
config-header    copyright     default
deprecated       descrip       detail
disable          documentation eaddr
enable           enabled       environrc
equivalence      exit-name     explain
export           extract-code  field
file-fail-code   flag          flag-code
flag-proc        flags-cant    flags-must
full-usage       gnu-usage     guard-option-names
help-value       homerc        ifdef
ifndef           immed-disable immediate
include          lib-name      library
long-opts        main          main-text
main-type        max           min
more-help-value  must-set      name
no-command       no-libopts    no-misuse-usage
no-preset        no-xlate      nomem-fail-code
omitted-usage    package       prefix
prefix-enum      preserve-case prog-name
prog-title       reorder-args  resettable
scaled           settable      short-usage
stack-arg        std-value     test-main
translators      unstack-arg   usage
usage-message    usage-opt     usage-type
val-name         val-upname    value
version
@end example

@node Option Define Names
@section Option Definition Name Index
@printindex vr

@ignore
END   == AUTOINFO == DO NOT CHANGE THIS COMMENT or the surrounding 'ignore's
Extraction from autogen.texi
@end ignore

@c = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
@c
@c  TRAILER

@c  LocalWords:  AutoGen texinfo Korb tpl bruce Exp texi autogen setfilename AG
@c  LocalWords:  settitle setchapternewpage dne dircategory direntry ifinfo gpl
@c  LocalWords:  AutoOpts snprintfv titlepage vskip pt filll sp dir xref cindex
@c  LocalWords:  AutoGen's noindent rc ini enum IDX const az upcase ENDFOR ESAC
@c  LocalWords:  optargs egcs inclhack sh fixincl autoconf endif var templ dirs
@c  LocalWords:  def txt cd STR str ifdef alist downcase sprintf arg lexer
@c  LocalWords:  srcfile linenum subblock defParse srcdir sed POSIX printf expr
@c  LocalWords:  stdout expr func gfunc tr findex exparg desc desc sep macfunc
@c  LocalWords:  ing getdefs libopts src ksh forcomma csh env Sourced autoopts
@c  LocalWords:  mkmerge builddir ADDON AutoGetopts getopt glibc argp perl awk
@c  LocalWords:  printindex cp fn