This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
H.Merijn Brand [Tue, 22 Sep 2020 11:17:11 +0000 (13:17 +0200)]
Trap SIGBUS when Configure checks for va_copy (#18148)
On several systems the attempt to determine if we "need va_copy() or
similar" results in a SIGBUS instead of a SIGSEGV because the fault
happens pointing at an io device address.
While reviewing accounting logs during the OpenBSD system build, bluhm@
noticed a core dump and deraadt@ tracked it down.
While researching it was discovered that this is a fairly common case in
other systems as well.
Original diff by deraadt@
Sawyer X [Mon, 21 Sep 2020 18:00:20 +0000 (20:00 +0200)]
Update epigraphs with links
Sawyer X [Mon, 21 Sep 2020 17:58:12 +0000 (19:58 +0200)]
Bump Module::CoreList in Maintainers.pl
Sawyer X [Sun, 20 Sep 2020 23:56:46 +0000 (01:56 +0200)]
Update perldelta with Module::CoreList
Sawyer X [Sun, 20 Sep 2020 23:56:34 +0000 (01:56 +0200)]
Update Module::CoreList for 5.33.3
Sawyer X [Sun, 20 Sep 2020 23:48:36 +0000 (01:48 +0200)]
Bump the perl version in various places for 5.33.3
Sawyer X [Sun, 20 Sep 2020 23:28:27 +0000 (01:28 +0200)]
new perldelta for 5.33.3
Sawyer X [Sun, 20 Sep 2020 23:17:46 +0000 (01:17 +0200)]
Tick off release
Sawyer X [Sun, 20 Sep 2020 23:10:57 +0000 (01:10 +0200)]
Add epigraph
Sawyer X [Sun, 20 Sep 2020 12:09:32 +0000 (14:09 +0200)]
add new release to perlhist
Sawyer X [Sun, 20 Sep 2020 11:40:58 +0000 (13:40 +0200)]
Update perldelta
Sawyer X [Sun, 20 Sep 2020 11:31:04 +0000 (13:31 +0200)]
Update Module::CoreList for 5.33.2
Sawyer X [Sun, 20 Sep 2020 09:15:46 +0000 (11:15 +0200)]
Update authors
Sawyer X [Sun, 20 Sep 2020 09:06:41 +0000 (11:06 +0200)]
Clean up perldelta
Sawyer X [Sun, 20 Sep 2020 08:21:22 +0000 (10:21 +0200)]
Doing the release today
Todd Rinaldo [Fri, 18 Sep 2020 23:07:02 +0000 (18:07 -0500)]
Update porting/customized.dat for recent change to cpan/ExtUtils-PL2Bat/t/make_executable.t
Todd Rinaldo [Fri, 18 Sep 2020 22:07:00 +0000 (17:07 -0500)]
Mark cpan/ExtUtils-PL2Bat/t/make_executable.t as customized
Goes with #18118. Per ilmari, this is ok to push in while we wait
for https://github.com/Perl-Toolchain-Gang/extutils-pl2bat/pull/6
to merge.
Christian Walde [Thu, 10 Sep 2020 22:24:46 +0000 (00:24 +0200)]
make_executable.t: ensure PATH for test is not too long, as that is incompatible with win2k
Hugo van der Sanden [Thu, 17 Sep 2020 14:28:42 +0000 (15:28 +0100)]
embed.fnc docfix: s/s/S/
Christian Walde [Wed, 16 Sep 2020 19:06:48 +0000 (21:06 +0200)]
prevent a spurious error output from test.pl when killing a cygwin process
Under Cygwin a process can sometimes take a little while to spool down
after being killed. There already is code to wait a second and retry.
However if the process has already disappeared in the wait second, then the
retry is engaged anyhow and will then complain it can't find the process.
This change makes it so test.pl only truly attempts to kill a cygwin
process if it actually is still around.
This resolves the secondary bug in #18129.
Nicholas Clark [Wed, 16 Sep 2020 17:38:17 +0000 (17:38 +0000)]
Skip t/porting/bench.t if the cachegrind test takes too long.
On this armv7l system (current Raspberry Pi OS) /usr/bin/valgrind exists,
but if I try to run it it gets stuck in a CPU busy loop somewhere. As was,
this meant that t/porting/bench.t would never complete, causing the test
suite to hang.
As we are already assuming ulimit exists to suppress coredumps, also use it
to set a CPU time limit. This way, the sanity check will time out (and fail)
causing t/porting/bench.t to skip, and the regression tests to pass.
On other systems the generous timeout is harmless, and the test still runs as
intended.
Todd Rinaldo [Wed, 16 Sep 2020 15:01:31 +0000 (10:01 -0500)]
Update DB_File to CPAN version 1.854
[DELTA]
1.854 16 September 2020
* Prefer direct notation over indirect (#4)
* Make hint/BS snippets strict compliant.
* trim whitespace
James E Keenan [Wed, 16 Sep 2020 00:45:03 +0000 (00:45 +0000)]
Revise per feedback
James E Keenan [Tue, 8 Sep 2020 00:48:01 +0000 (00:48 +0000)]
t/op/taint.t: Avert uninitialized value warning
In the test of shmread, if $id is undefined after the attempted
assignment by shmget(), we get the "# shmget failed" warning and, since
nothing has been read into $rcvd, that variable remains undefined. If
$rcvd is undefined it will never be string-equal to $sent, which
means the 'skip' condition will be exercised.
Commit
56249513ac made this file run with warnings enabled, so now the
lack of assignment to $rcvd generates an uninitialized value warning
(observed in smoke-test reports on FreeBSD). This patch accounts for
non-definition of $rcvd and prevents that warning from being emitted.
Todd Rinaldo [Wed, 16 Sep 2020 02:23:31 +0000 (21:23 -0500)]
Update Test-Simple to CPAN version 1.302181
[DELTA]
1.302181 2020-09-14 09:46:04-07:00 America/Los_Angeles
- put try_sig_mask back where it goes (And add test to prevent this in the future)
- Drop new List::Util requirement back down
1.302180 2020-09-13 23:11:18-07:00 America/Los_Angeles
- No changes since last trial
1.302179 2020-09-12 22:35:19-07:00 America/Los_Angeles (TRIAL RELEASE)
- Bump minimum List::Util version (for uniq)
1.302178 2020-09-07 14:11:52-07:00 America/Los_Angeles (TRIAL RELEASE)
- Move try_sig_mask to the only module that uses it.
- Inherit warnings bitmask in cmp_ok string eval
- Update copyright date
- Improved API for interept {} and what it returns
1.302177 2020-08-06 21:46:06-07:00 America/Los_Angeles
- Minor fix to author downstream test
- No significant changes since the last trial
1.302176 2020-08-05 21:45:19-07:00 America/Los_Angeles (TRIAL RELEASE)
- Fix Test::More's $TODO inside intercept (#862)
Nicolas R [Tue, 1 Sep 2020 22:10:25 +0000 (16:10 -0600)]
fix t/win32/signal.t
Make sure Config is loaded before using %Config
which is required when using 'strict vars'.
Todd Rinaldo [Tue, 15 Sep 2020 23:42:49 +0000 (18:42 -0500)]
Update ExtUtils-Install to CPAN version 2.18
[DELTA]
2.18
- Add descriptions for 3 tests lacking them
- Removed bundled Test::More from t/lib
- 'use strict' added to all files where missing
- 'use vars' replaced with 'our'
- Properly scope all variables
- Minimal supported perl version is now v5.6.0
Todd Rinaldo [Tue, 15 Sep 2020 22:40:55 +0000 (17:40 -0500)]
Update IO-Socket-IP to CPAN version 0.41
[DELTA]
0.41 2020-09-15 16:59
[CHANGES]
* Updated to `=head2 barename` documentation style
[BUGFIXES]
* Remember to declare configure_requires Module::Build
0.40 2020-09-15 16:41
[CHANGES]
* Don't ship a Makefile.PL
* Updated docs about `-register` option
[BUGFIXES]
* Ensure safety under default-strict (RT133322)
Todd Rinaldo [Tue, 15 Sep 2020 14:43:03 +0000 (09:43 -0500)]
Todd Rinaldo [Tue, 15 Sep 2020 14:37:28 +0000 (09:37 -0500)]
Update Compress-Raw-Zlib to CPAN version 2.096
[DELTA]
2.096 31 July 2020
* No changes
2.095 19 July 2020
* No changes
2.094 13 July 2020
* Issue with Append mode & SvOOK
https://github.com/pmqs/Compress-Raw-Zlib/issues/3
80ee0524012f46c5984c2d57649af0b07f82c750
Todd Rinaldo [Tue, 15 Sep 2020 14:23:37 +0000 (09:23 -0500)]
Update Compress-Raw-Bzip2 to CPAN version 2.096
Address .gitignore issue with *.patch by making it /*.patch. This
seems to match the original intention of the ignore. This also
corrects a mainfest failure because bzip2-src now includes a .patch
file.
[DELTA]
2.096 31 July 2020
* No changes
2.095 19 July 2020
* No changes
2.094 13 July 2020
* Issue with Append mode & SvOOK
https://github.com/pmqs/Compress-Raw-Bzip2/issues/2
e6f1fa7d86a4c631b82b7c923a5def3e274925fa
Todd Rinaldo [Sun, 13 Sep 2020 18:05:29 +0000 (13:05 -0500)]
Non-Configure code required to implement a strict by default option
Todd Rinaldo [Sun, 13 Sep 2020 18:02:25 +0000 (13:02 -0500)]
Add the Configure option to enable strict by default in perl programs.
Do not set strict by default on -e or -E
This is a development only feature to allow us to clean up blead.
./Configure -Dusedevel -des -Dusedefaultstrict
Hugo van der Sanden [Tue, 15 Sep 2020 14:26:51 +0000 (15:26 +0100)]
Update a couple of RT references in todo.pod
Nicolas R [Mon, 14 Sep 2020 17:23:11 +0000 (11:23 -0600)]
Update IO-Zlib to CPAN version 1.11
Enable strict and warnings.
Nicolas R [Mon, 3 Aug 2020 04:34:35 +0000 (22:34 -0600)]
Fix Win32 errors from Errno
Fix #18025
This is fixing these two errors under windows:
"Never use <avx512bwintrin.h> directly; include <immintrin.h> instead."
"Never use <avx512vpopcntdqvlintrin.h> directly; include <immintrin.h> instead."
Karl Williamson [Sun, 6 Sep 2020 03:16:04 +0000 (21:16 -0600)]
perlapi: Consolidate sv_utf8_downgrade-ish entries
Karl Williamson [Mon, 14 Sep 2020 04:08:47 +0000 (22:08 -0600)]
perlapi: Fix function prototypes
These were wrong, and Devel::PPPort found the mistakes.
Karl Williamson [Mon, 14 Sep 2020 04:05:42 +0000 (22:05 -0600)]
perlapi: Fix typo.
Karl Williamson [Mon, 14 Sep 2020 04:03:39 +0000 (22:03 -0600)]
Fix input type on isFOO macros
These macros accept any UV as input, but not a signed number
Karl Williamson [Mon, 14 Sep 2020 15:30:38 +0000 (09:30 -0600)]
podcheck db: Note that atof(3) exists
Karl Williamson [Mon, 14 Sep 2020 04:02:10 +0000 (22:02 -0600)]
gv_fetchpvn and gv_fetchpvn_flags are identical
Karl Williamson [Wed, 22 Jul 2020 22:22:27 +0000 (16:22 -0600)]
Document my_atof
Karl Williamson [Wed, 22 Jul 2020 12:28:36 +0000 (06:28 -0600)]
embed.fnc: Mark str_to_version as Core only
This is an internal toker function, with no uses on cpan
Karl Williamson [Tue, 8 Sep 2020 17:28:37 +0000 (11:28 -0600)]
perlguts: Remove redundant entries
I didn't get everything cleaned up in earlier commits
Karl Williamson [Wed, 22 Jul 2020 12:10:54 +0000 (06:10 -0600)]
Document cast NV to int macros; make helpers internal
The helper, like, cast_uv, functions are now considered non-API, but the
macros one is supposed to use instead are now documented.
Karl Williamson [Sat, 29 Aug 2020 03:18:41 +0000 (21:18 -0600)]
Document AvFILLp; clarify related entries
Karl Williamson [Mon, 14 Sep 2020 03:03:18 +0000 (21:03 -0600)]
Fix podcheck.t data.
This commit does what
909372cc412056a2ced94ef28c651110a788b6ff was
intended to do, but I screwed up.
Karl Williamson [Sat, 29 Aug 2020 19:55:13 +0000 (13:55 -0600)]
perlapi: Slight improvement to sv_2pvbyte
Todd Rinaldo [Sun, 13 Sep 2020 04:45:25 +0000 (23:45 -0500)]
Update Module-Load-Conditional to CPAN version 0.74
Ignore that versions have not changed in test file libraries.
[DELTA]
0.74 Fri Aug 21 09:06:21 BST 2020
* strictify the test suite packages
0.72 Thu 25 Jun 08:21:16 BST 2020
* Avoid using indirect calls
Todd Rinaldo [Sun, 13 Sep 2020 03:38:27 +0000 (22:38 -0500)]
Update Filter-Util-Call to CPAN version 1.60
[DELTA]
1.60 2020-08-05 rurban
- Increase t/call.t verbosity on failures (PR #12 aatomic)
- Push cwd to @INC for PERL_CORE (PR #11 jkeenan)
- Update search.cpan.org link to metacpan (PR #10 Grinnz)
Todd Rinaldo [Sat, 12 Sep 2020 03:55:15 +0000 (22:55 -0500)]
Update Archive-Tar to CPAN version 2.38
[DELTA]
2.38 25/06/2020 (ATOOMIC)
- Avoid indirect calls
- Add use warnings to bin/ptar*
Craig A. Berry [Fri, 11 Sep 2020 19:31:08 +0000 (14:31 -0500)]
.PHONY is specific to Unix make utilities.
And in MMS or MMK on VMS it's a syntax error. So only spell it
with a dot when not using those utilities to run Devel::PPPort's
Makefile.PL.
John Lightsey [Tue, 25 Aug 2020 21:00:52 +0000 (16:00 -0500)]
Make security warnings in the Safe and Opcode modules more direct.
The current "no warranty" text warning against the use of Safe or
Opcode for "security purposes" is somewhat ambiguous. These modules
are not effective sandboxing mechanisms for evaluating untrusted
perl code and should not be used in that manner.
Safe and Opcode are, at best, hardening measures that could be used
in combination with operating system level sandboxing of the perl
interpreter.
John Lightsey [Tue, 25 Aug 2020 20:49:05 +0000 (15:49 -0500)]
Document the security team's processes and practices
Add a new perlsecpolicy POD file with detailed descriptions
of the security team's vulnerability remediation workflow and
the criteria used to distinguish security issues from other
types of bugs.
This also switches the team's public contact address to
perl-security@perl.org, and updates the security contact
information shown in github's issue interface.
Karl Williamson [Sun, 6 Sep 2020 03:45:54 +0000 (21:45 -0600)]
perlapi: Consolidate sv_dec-ish entries
Karl Williamson [Sun, 6 Sep 2020 19:10:02 +0000 (13:10 -0600)]
Use macro instead of its expansion
We already have a macro that expands to what this code does; it's
clearer to use it.
Karl Williamson [Sat, 5 Sep 2020 18:13:50 +0000 (12:13 -0600)]
perlapi: Consolidate sv_set[inu]v-ish entries
Karl Williamson [Sun, 6 Sep 2020 18:52:28 +0000 (12:52 -0600)]
perlapi: Fix poor wording in MUTABLE_PTR
H.Merijn Brand [Mon, 7 Sep 2020 12:26:35 +0000 (14:26 +0200)]
Missing update
I need this to auto-happen :)
H.Merijn Brand [Mon, 7 Sep 2020 12:24:28 +0000 (14:24 +0200)]
Comment for L_R_TZSET now correctly added to config_h.SH
Karl Williamson [Mon, 7 Sep 2020 02:26:08 +0000 (20:26 -0600)]
regen podguts after mg_vtable.pl changes
Caused in
eb0444cb23a8ec47ef18d5aa33ec88aa5e71b7e8
Ricardo Signes [Mon, 7 Sep 2020 00:38:52 +0000 (20:38 -0400)]
perlop: put the stringy bitwise ops in precedence table
Ricardo Signes [Mon, 7 Sep 2020 00:30:36 +0000 (20:30 -0400)]
perldesc: replace circumfix with postfix dereference
...and what an odd one, this!
Ricardo Signes [Mon, 7 Sep 2020 00:29:53 +0000 (20:29 -0400)]
perldsc: eliminate unnecessary call to keys()
A hash in scalar context now evalutes to its key count, meaning that
(%h1 <=> %h2) is equivalent to, but shorter than, (keys %h1 <=> keys
%h2). I have dropped calls to keys() in this code to make it clearer.
(This commit may be controversial, as in some places scalar(@arr) is
being used with the suggestion that it provides values.)
Ricardo Signes [Mon, 7 Sep 2020 00:28:20 +0000 (20:28 -0400)]
perldsc: slightly tweak examples for clarity
Ricardo Signes [Mon, 7 Sep 2020 00:28:02 +0000 (20:28 -0400)]
perldsc: remove suggestion of more docs coming someday
Ricardo Signes [Mon, 7 Sep 2020 00:27:40 +0000 (20:27 -0400)]
perldsc: remove the reference to using -w
Ricardo Signes [Mon, 7 Sep 2020 00:25:23 +0000 (20:25 -0400)]
perldsc: in general, prefer postfix to circumfix dereference
In all (or almost all) cases, postfix dereferencing is simpler to read
and write. This makes it the better choice for teaching new Perl
programmers.
I have left in reference to circumfix deref, but mostly to acknowledge
that it exists, not to suggest its use.
Ricardo Signes [Mon, 24 Aug 2020 01:30:35 +0000 (21:30 -0400)]
perlretut: make example code pass under strict
Ricardo Signes [Mon, 24 Aug 2020 01:30:21 +0000 (21:30 -0400)]
language docs: replace one pointless use of circumfix deref
Ricardo Signes [Mon, 24 Aug 2020 01:30:06 +0000 (21:30 -0400)]
language docs: prefer postfix deref in numerous places
Karl Williamson [Mon, 7 Sep 2020 02:26:08 +0000 (20:26 -0600)]
regen/mg_vtable.pl: Fix perlapidoc display
This wasn't generating the right thing
Karl Williamson [Sun, 6 Sep 2020 00:04:39 +0000 (18:04 -0600)]
perlapi: Clarify croak_no_modify()
Karl Williamson [Sat, 5 Sep 2020 23:51:45 +0000 (17:51 -0600)]
perlapi: Consolidate some nocontext fcns with their main fcn
Karl Williamson [Thu, 18 Jun 2020 09:05:37 +0000 (03:05 -0600)]
mg.c: Use SvTRUE instead of sv_true
The former is preferred. This was the only sv_true() use in the core.
Karl Williamson [Tue, 16 Jun 2020 22:42:47 +0000 (16:42 -0600)]
Clarify branch prediction in SvTRUE
Karl Williamson [Sat, 29 Aug 2020 19:35:29 +0000 (13:35 -0600)]
embed.fnc: Mark pad_setsv as core-only
You are supposed to access this functionality through PAD_SETSV
Karl Williamson [Thu, 27 Aug 2020 21:22:22 +0000 (15:22 -0600)]
perlapi: Clarify SV Handling
Karl Williamson [Thu, 27 Aug 2020 20:05:19 +0000 (14:05 -0600)]
perlapi: Display PERL_VERSION_xy better
Now that we have the tools to show these as a whole list, do so
Karl Williamson [Thu, 27 Aug 2020 15:42:01 +0000 (09:42 -0600)]
Document mini_mktime
Karl Williamson [Thu, 27 Aug 2020 13:41:39 +0000 (07:41 -0600)]
Document INT16_C, etc
Karl Williamson [Thu, 27 Aug 2020 12:58:25 +0000 (06:58 -0600)]
Change docs display for PERL_UNUSED_foo
These should have a semicolon after each in their usage
Karl Williamson [Thu, 27 Aug 2020 12:57:28 +0000 (06:57 -0600)]
Document PERL_UNUSED_DECL
Karl Williamson [Thu, 27 Aug 2020 12:56:05 +0000 (06:56 -0600)]
Document cPERLscope
Karl Williamson [Wed, 26 Aug 2020 22:46:16 +0000 (16:46 -0600)]
README.os2: Fix pod link typo
Karl Williamson [Wed, 22 Jul 2020 22:43:11 +0000 (16:43 -0600)]
Remove redundant documentation
Now that we can get automatic links in perlapi, remove the redundant
stripped-down documentation that goes there in favor of the better
documentation in perlcall.
Karl Williamson [Wed, 22 Jul 2020 23:18:31 +0000 (17:18 -0600)]
embed.fnc: Add text to comment
Karl Williamson [Wed, 22 Jul 2020 22:46:57 +0000 (16:46 -0600)]
Document vload_module
Karl Williamson [Wed, 22 Jul 2020 22:07:45 +0000 (16:07 -0600)]
Note that some dump constant flags are not public
One of these is no longer used; the others have synonyms that are
public. None are used in cpan
Karl Williamson [Wed, 22 Jul 2020 12:34:35 +0000 (06:34 -0600)]
embed.fnc: Mark several regex helpers as Core only
These are for internal regex use.
Karl Williamson [Sat, 18 Jul 2020 13:27:48 +0000 (07:27 -0600)]
Document several warner() fcns
Karl Williamson [Sat, 5 Sep 2020 22:40:17 +0000 (16:40 -0600)]
autodoc: Avoid a loop iteration
Initializing everything with the first elements values allows us to not
look at that element again. Previously, only somethings were so
initialized.
Karl Williamson [Sat, 5 Sep 2020 22:37:22 +0000 (16:37 -0600)]
autodoc: Properly take into account pTHX
Before this patch, it wasn't considering the thread context when
deciding whether two forms of similar functions had the same signature
or not; hence not displaying things when it should have.
Karl Williamson [Sat, 5 Sep 2020 21:30:25 +0000 (15:30 -0600)]
mg.h: White-space only
Karl Williamson [Sat, 5 Sep 2020 21:29:57 +0000 (15:29 -0600)]
Document whichsig and kin
Karl Williamson [Sat, 5 Sep 2020 21:19:46 +0000 (15:19 -0600)]
autodoc: Correctly display some function signatures
The code wasn't properly showing signatures when the function was a
subsidiary one in an entry containing multiple functions, and the
function's signature was specified in embed.fnc
Karl Williamson [Sat, 5 Sep 2020 21:35:22 +0000 (15:35 -0600)]
porting/known_pod_issues.dat: Note vsprintf() exists
Karl Williamson [Sat, 5 Sep 2020 18:01:51 +0000 (12:01 -0600)]
Document sv_vcatpvfn
Consolidate it with sv_vcatpvfn_flags
Karl Williamson [Sat, 5 Sep 2020 15:39:38 +0000 (09:39 -0600)]
autodoc: Fix for win32.
This platform has config.h located in a different place
Thanks to Christian Walde for testing this