| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Update the checks for libjpeg to try libjpeg.pc with
PKG_CHECK_MODULES first, and also to allow explicitly
set LIBJPEG_CFLAGS and LIBJPEG_LIBS.
The old detection logic did not allow specifying the location
of libjpeg at all. It only allowed either disabling libjpeg
or using libjpeg from the default include and linker paths.
So this also changes the places in the Makefiles where we
used to add @LIBJPEG@ to the linker to both link with
$(LIBJPEG_LIBS) and compile with $(LIBJPEG_CFLAGS).
The new configure logic keeps --without-jpeg to force-
disable libjpeg, but updates the detection of libjpeg
to first try the LIBJPEG_* variables if given, otherwise
try PKG_CHECK_MODULES for libjpeg, and only then falls
back to the old way of just trying the default include
and link locations for libjpeg.
This mentions more information in the configure summary
if not building with libjpeg.
This is the first time we make it possible to build with
libjpeg from e.g. FreeBSD ports' or MacOS homebrew's
jpeg-turbo packages.
|
|
|
|
|
|
|
|
|
|
| |
We still do not use LTDL_INIT() as LTDL_INIT insists on
us shipping libltdl sources in our release tarballs and
we still do not like shipping third party sources.
This now checks locations other than the default include
and linker paths, so that configure should find libltdl
by itself now on e.g. FreeBSD or OSX/homebrew.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Make the compiler warnings work with clang in addition to gcc by
test compiling all warning options before actually using them.
This came up when the pedantic build tests failed on FreeBSD 13.
To achieve this, we add -Werror=unknown-warning-option to
the CFLAGS for clang to abort immediately on encountering
an unknown warning option. This means we now only add warnings
which the compiler (clang, gcc, or whatever) actually supports.
As running configure now checks whether compiling is possible
for each warning flag, this should not introduce compatibility
issues where builds which used to work now fail to build.
However, you can always opt to not use the generated CFLAGS
by setting CFLAGS to empty or whatever else you want when
running your build, e.g. like
make all check CFLAGS=
|
|
|
|
|
|
|
|
| |
Add a pedantic C compile test for c2x in addition to
the existing ones for c90, c99, and c11.
Add a pedantic C++ compile test for c++20 in addition to
the existing ones for c++98, c++11, c++14, and c++17.
|
|
|
|
|
|
|
| |
Always define the test programs, but only conditionally test them.
This makes certain that they can be build, that the sources are
added to the dist tarball, etc.
|
|
|
|
|
|
|
|
|
|
|
|
| |
These did not work with C++ on FreeBSD 13 anyway, due to
c++98 not working there, which the old pedantic compilation
just assumed was working without testing it in configure.
The new pedantic compilation tests offer a multitude of
language standards including c++98, but only tries building
the pedantic compilation tests for those language standards
for which configure has shown that building programs is
possible at all.
|
|
|
|
|
|
|
|
| |
Fix the GP_SLEEP invocation in the Gentoo hack warning which
adds the ptp2 camlib if the only camlib requested is canon, as
modern Canon cameras use the ptp2 camlib, not the canon camlib.
Also syncs gphoto-m4/gp-camlibs.m4 to libgphoto2-port/gphoto-m4/.
|
|
|
|
|
|
| |
The printf(3) format string character "%m" is available on
Linux/glibc and FreeBSD, but not on MacOSX. So using "%m"
should be limited to system specific code.
|
| |
|
| |
|
|
|
|
|
|
|
| |
We have been requiring at least parts of C99 like stdint.h for some
time now, but HACKING.md has not reflected that until now.
Closes: https://github.com/gphoto/libgphoto2/issues/724
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
Declare local variable local by declaring it static.
This avoids the clang warning about the variable not being declared
static.
|
|
|
|
|
| |
In the pedantic compilation tests, include the system headers
before gphoto2 headers - which should be the way to do things.
|
|
|
|
|
|
|
| |
Add remaining source files including libgphoto2/i18n.h
or libgphoto2_port/i18n.h to the respective POTFILES.in,
regardless of whether or not the source file actually
contains translated strings.
|
|
|
|
|
|
|
| |
All three added sourcefiles had the i18n headers,
but strings actually marked for translations are
only found in ptpip/ptpip.c, not in any of the
vusb/vcamera.c or vusb/vusb.c files.
|
|
|
|
|
| |
Rename the libgphoto2_port/test directory to libgphoto2_port/tests
to make its name the same as libgphoto2/tests.
|
| |
|
| |
|
|
|
|
| |
Delete some old debug comments from the configure script.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Factor out the i18n definitions into i18n.h include files:
One include file each for libgphoto2 and for libgphoto2_port.
This removes the complete
#ifdef ENABLE_NLS
# include <libintl.h>
# undef _
# define _(String) dgettext (GETTEXT_PACKAGE, String)
# ifdef gettext_noop
# define N_(String) gettext_noop (String)
# else
# define N_(String) (String)
# endif
#else
# define _(String) (String)
# define N_(String) (String)
#endif
block (which in some occasions defines even more macros)
and replaces it with a single line
#include "libgphoto2/i18n.h"
for camlibs and libgphoto2 itself and with
#include "libgphoto2_port/i18n.h"
for iolibs and libgphoto2_port itself.
This gives us two central locations to change the defintions.
|
|
|
|
|
|
|
|
|
|
|
| |
The gphoto-m4 gp-check-library.m4 has been using PKG_PROG_PKG_CONFIG
since commit c1ab67b7288359066b4884d3058e712f6170da35 (2017-08-29),
so this makes the libgphoto2_port/gphoto-m4 use the same code.
As PKG_PROG_PKG_CONFIG provided by pkg-config's pkg.po, this means
we now require that to be present - and we should better check at
autoreconf time that the PKG_PROG_PKG_CONFIG macro is actually
defined.
|
|
|
|
|
|
|
|
| |
Using EOF does the job, is well readable, and there is
just no reason to use any other string.
On the contrary, it may cause additional trouble when
using m4_pattern_forbid.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
see https://github.com/gphoto/libgphoto2/pull/738
|
| |
|
|
|
|
| |
Use it in ptp2 PTP_EC_ObjectInfoChanged.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
As the German translation ("de") is configured as "external" at
the Translation Project, there will be no updated `de.po` files
from the Translation Project to overwrite what we have in our
source tree.
So the gphoto project needs to maintain the de.po file independent
from the Translation Project, and updating the file from ISO-8859-1
to UTF-8 is a part of that.
|
|
|
|
| |
e.g. if you select "protect" on the camera.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The linux-hotplug mechanism has been replaced by udev and hwdb for
quite some time now, so we can remove the support for packaging
libgphoto2 with linux-hotplug.
For a rough idea of the timeline of linux-hotplug being replaced,
the last release of hotplug appears to have been in March of 2004:
https://sourceforge.net/projects/linux-hotplug/files/
and the Fedora 5 (released March 20, 2006) release notes
mention the hotplug package having been removed.
|
|
|
|
|
|
| |
There are no *.sh files being generated by AC_CONFIG_FILES,
so the code bit setting the executable bit on *.sh is not
needed.
|
|
|
|
|
|
|
|
|
|
| |
While adding support for "make rpm" appeared to be a good idea in the
early 2000s, it really is not and only clutters up the build system.
At least the libgphoto2_port translation cannot have worked in the
rpm package since at least 5255d802102c10f971dc98f10828451d9922340d
from 2007-10-28, so it is a good guess that this has not really been
in use.
|
|
|
|
|
| |
For portability, use && and || instead of the test arguments
-a and -o.
|
| |
|
|
|
|
|
|
| |
Not that its presence hurt anything, but we could
have removed the libltdlc.la stuff some time in
the 2000s.
|
|
|
|
| |
see https://github.com/gphoto/gphoto2/issues/459
|