| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
My perldoc caused these split lines to be output wrongly.
|
|
|
|
|
| |
I looked at the differences of all pods in /pod, and found a bunch of
ones that I think are significant that weren't listed in perldelta.
|
|
|
|
|
|
|
|
|
|
|
| |
SCF_DO_SUBSTR is a flag internal to the current implementation of the
regular expression optimizer. There is no need to proclaim its
existence to the outside world, and is just extraneous noise.
I myself do not understand this flag, and I've spent more time looking
at this code than all but a few people likely to be reading this
perldelta. If someone who does understand it could explain it to me, I
would add comments to the code (after the freeze) to aid future readers.
|
| |
|
|
|
|
|
| |
Except when referring to actual names of things.
Also update the diagnostic description in perldiag.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Perl on MSVC6 doesnt support 64 bit ints (p5p choice not to support it)
so this macro isn't defined on MSVC6 builds, commit e59642234e hid this
mistake from non-DEBUGGING builds. The mistake is a copy paste mistake
from commit eacbb37937 . Miniperl fails at VC6 link time due to UINT64_C
symboil not being found.
..\pad.c(165) : warning C4013: 'UINT64_C' undefined; assuming extern
returning int
..\pad.c(165) : warning C4018: '!=' : signed/unsigned mismatch
|
|
|
|
|
|
|
|
|
|
|
| |
There are suspicions that the process tree kill('-KILL', ...) might be
nuking too much. It was only done to kill the cmd.exe+perl.exe tree that
was unexpectedly launched by system(1, $cmd), but by switching to
system({$perl} 1, $perl, '-e', $prog) we can avoid the intermediate cmd.exe
and thus revert to normal process kill('KILL', ...) instead to kill the
perl.exe that is now launched directly.
See http://www.nntp.perl.org/group/perl.perl5.porters/2015/05/msg227859.html
|
| |
|
| |
|
| |
|
|
|
|
|
| |
We may yet restore a summary of important changes, but for now,
this is it.
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 50e5165b9638b94be310f15477b42935c79e82d5.
That commit fixed the leak too well and instead introduced a potential
premature free.
This re-introduces the long-standing leak, which will be addressed post
5.22 release.
See RT #124181
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
These are my changes from between
=head1 Selected Bug Fixes
inclusive up to EOF.
Some notes.
The C<$lexical = chr $foo> optimisations were added and then removed
during 5.21.x, so I've removed all references to them.
Ditto readline after <<>> was added.
Ditto "Only stringy classnames are now shared".
Ditto "two assertion failures introduced into C<-DPERL_OP_PARENT>"
The size overflow stuff of 123554 was mentioned in two parts; now
consolidated into one.
Moved the two flip-flop items to be next to each other.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
These are my changes from between
=head1 Modules and Pragmata
inclusive up to (but not including)
=head1 Selected Bug Fixes
Some specific comments:
I moved "Missing or undefined argument to require" from "changed diags"
to "new diags": the behaviour has changed, but its a completely new diag
message.
"<B> now compiles again on Windows" I assume was only broken within the
5.21.x series.
The headline description of Perl_sv_get_backrefs() made it sound like it
returns a boolean. The more detailed description I've deleted, since
people can get that from the api doc.
I've merged the two entries for cv_name.
|
| |
|
|
|
|
| |
with blead
|
|
|
|
| |
See http://nntp.perl.org/group/perl.perl5.porters/228010
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
v5.19.3-139-g2537512 changed POPSUB and POPFORMAT so that they also
unwind the relevant portion of the scope stack. This (sensible) change
means that during exception handling, contexts and savestack frames are
popped in lock-step, rather than all the contexts being popped followed by
all the savestack contents.
However, LEAVE_SCOPE() is now called by POPSUB/FORMAT, which can trigger
destructors, tied method calls etc, which themselves may croak. The new
unwinding will see the old sub context still on the context stack and call
POPSUB on it again, leading to double frees etc.
At this late stage in code freeze, the least invasive change is to
use an unused bit in cx->blk_u16 to indicate that POPSUB has already
been called on this context frame.
Sometime later, this whole area of code really needs a thorough overhaul.
The main issue is that if cxstack_ix-- is done too early, then calling
destructors etc can overwrite the current context frame while we're still
using using it; if cxstack_ix-- is done too late, then that stack frame
can end up getting unwound twice.
|
| |
|
|\ |
|
| |
| |
| |
| | |
and the customized housekeeping to match
|
| | |
|
|/ |
|
|
|
|
|
|
| |
This borrows heavily from Barbie's longer obituary:
http://blogs.perl.org/users/barbie/2015/05/brian-mccauley-1967-2015.html
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
These are my changes up to (but not including)
=head1 Modules and Pragmata
I've expanded on things mainly where I personally was confused by what the
entry was trying to say, and where I had to refer back to the src or RT
ticket.
Some specific stuff:
"qr/foo/x> now ignores any Unicode pattern white space" appeared under
both "Core Enhancements" and "Incompatible Changes"; I've consolidated
both entries into one, in the latter section.
\C now in fact warns; it was already deprecated in an earlier release.
Performance efactoring of pp_tied and pp_ref is really a minor internal
detail not significant enough for a perldelta.
"Pathtools doesn't try to load XS on miniperl" - I've moved this from
"performance" to "build", since it doesn't affect the performance of perl
binary, but only how long it takes to build that binary.
"A typo fix reduces the size of the C<< OP >> structure." - this both
broke and was fixed within 5.21.x, so I've removed the entry.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Based on feedback by Shlomi Fish.
|
|
|
|
| |
links to external modules fail t/porting/podcheck.t unless exempted.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I added XXX to lines where the text for this perldelta look like it
ought to change somewhat due to this being the delta from 5.20.2 to
5.22.0, instead of from one 5.21 release to another. I believe, but am
uncertain that several of these should just be deleted; while others
require rewording. Rather than not call out attention to them, I am
marking them to draw attention to them for people more expert in these
areas to adddress.
Several areas seemed clear to me that a bug introduced in some 5.21
version was fixed in a later 5.21 version, so has zero relevance for
making the jump from 5.20 to 5.22. I just deleted those.
And I had minor rewording in others, where it mentioned 5.21. We've
decided not to talk about development releases in the pods.
|
| |
|
| |
|
| |
|
| |
|
| |
|