Jesse Vincent [Tue, 13 Oct 2009 14:32:38 +0000]
Podify the social contract about contributed modules
Smylers [Tue, 13 Oct 2009 12:14:46 +0000]
perlvar $1 clarification
$1 is currently documented as being set by the "last pattern match".
But it is left alone by unsuccessful pattern match attempts (continuing
to hold a value from an earlier successful match).
Saying "last successful pattern match" clarifies this; it's also the
phrase used to document $&.
Second, the entry for $1 in perlvar doesn't actually contain the text
"$1" anywhere. As such, doing man perlvar then using /\$1 to search for
it in less doesn't locate it (though does match other places in that
file where $1 happens to be used).
Nicholas Clark [Tue, 13 Oct 2009 12:01:52 +0000]
Change S_ckwarn_common() to looping over the packed warning value.
The core never uses WARN3() or WARN4(), and rarely uses WARN2(), so the
previous code, effectively an unwrapped loop, wasn't a speed up. Functionally
equivalent smaller code fits better into CPU caches.
Nicholas Clark [Tue, 13 Oct 2009 10:41:36 +0000]
Migrate common code in Perl_ckwarn() and Perl_ckwarn_d() to S_ckwarn_common()
Nicholas Clark [Tue, 13 Oct 2009 10:30:14 +0000]
In Perl_ckwarn() and Perl_ckwarn_d() disentangle the complex conditionals.
It's much easier to see what is going on, if we use multiple return statements.
Nicholas Clark [Tue, 13 Oct 2009 10:05:49 +0000]
Replace Perl_deprecate() with a macro that calls Perl_ck_warner()
Perl_deprecate was not part of the public API, and did not have a deprecate()
shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
CPAN::Unpack show any users outside the core.
Nicholas Clark [Tue, 13 Oct 2009 09:27:36 +0000]
Move the code related to deprecate("comma-less variable list") into a function.
This makes all calls to deprecate() string constants.
Nicholas Clark [Tue, 13 Oct 2009 06:37:35 +0000]
Remove category 'syntax' from 5 warnings that should just be in 'deprecated'.
None were documented as also being in 'syntax'. Effectively, this completes the
reorganisation of commits
12bcd1a617c74d6e and 9014280dc8264580. See
http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2009-10/msg00601.html and
http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-03/msg00850.html
Nicholas Clark [Tue, 13 Oct 2009 06:12:04 +0000]
Use deprecate() in place of deprecate_old(), and remove S_deprecate_old().
Nicholas Clark [Mon, 12 Oct 2009 17:48:38 +0000]
ckWARNreg() and ckWARNregdep() now use string concatenation, like their brethren
They had been concatenating "%s" REPORT_LOCATION, as they weren't passing in a
format string, which wasn't consistent with the 2-5 argument versions. None of
the strings passed in have % characters in them, so this is safe (and any static
analyser will be able to see this).
Nicholas Clark [Mon, 12 Oct 2009 17:32:12 +0000]
Replace "%s"/SvPVX(sv_name) with "%"SVf/SVfARG(sv_name), which is UTF-8 safe.
Nicholas Clark [Mon, 12 Oct 2009 16:59:49 +0000]
In regcomp.c, where possible replace vWARN* with ckWARN*, using Perl_ck_warner.
This folds many pairs of ckWARN*() && Perl_warner() calls into singles call to
Perl_ck_warner(). vWARN(), vWARNdep() and vWARN2() are no longer used, so are
removed.
Geoffrey T. Dairiki [Wed, 5 Aug 2009 00:54:34 +0000]
overload no longer implicitly unsets fallback on repeated 'use overload' lines - Fix for RT#68916
Subject: overload::import resets the setting of 'fallback'
Date: Tue, 04 Aug 2009 17:54:34 -0700
From: "Geoffrey T. Dairiki" <dairiki at dairiki.org>
Steve Hay [Mon, 12 Oct 2009 16:21:05 +0000]
Remove Test-Simple's 00compile.t from core
It required a hand-edited version of Test-Simple's MANIFEST due to the
exclusion of a .pm file from core, but we shouldn't have edited versions
of CPAN files in core. The presence of the MANIFEST file also caused
warnings from the Makefile.PL due to other files being excluded from core,
so rather than further editing of the MANIFEST it is simplest to just
remove it and the test that requires it.
Nicholas Clark [Mon, 12 Oct 2009 15:39:02 +0000]
Add Perl_ck_warner_d(), which combines Perl_ckwarn_d() and Perl_warner().
Replace ckWARN_d{,2,3,4}() && Perl_warner() with it, which trades reduced code
size for 1 more function call if warnings are not enabled.
Nicholas Clark [Mon, 12 Oct 2009 15:06:20 +0000]
Add Perl_ck_warner(), which combines Perl_ckwarn() and Perl_warner().
Replace ckWARN{,2,3,4}() && Perl_warner() with it, which trades reduced code
size (about 0.2%), for 1 more function call if warnings are not enabled.
However, if we're now in the L1 or L2 cache when we weren't previously, that's
still going to be a speed win.
Nicholas Clark [Mon, 12 Oct 2009 14:54:30 +0000]
Normalise 3 fatal pack/unpack error messages to "panic: %s"
Nicholas Clark [Mon, 12 Oct 2009 14:28:29 +0000]
In uni_to_bytes(), fix ckWARN() anomaly.
Nicholas Clark [Mon, 12 Oct 2009 13:53:52 +0000]
Re-order a conditional with ckWARN() last, as it is a function call.
Steve Hay [Mon, 12 Oct 2009 10:43:08 +0000]
version's t/survey_locales is not in core so mark it as EXCLUDED
Nicholas Clark [Mon, 12 Oct 2009 10:24:15 +0000]
deprecate_old() is not public, and only used within toke.c, so can be static.
Nicholas Clark [Mon, 12 Oct 2009 10:03:42 +0000]
In Perl_ckwarn(), simplify the check of isLEXWARN_on/isLEXWARN_off.
Rafael Garcia-Suarez [Mon, 12 Oct 2009 08:30:42 +0000]
Don't install filter-util.pl
This file was getting incorrectly installed with perl since commit
2adbc9b6919cad1240a834325b355e9b2d5efd67, that moved it from t/ to ext/.
Steve Hay [Sun, 11 Oct 2009 23:30:06 +0000]
Implement TODO support locally in t/comp/opsubs.t
Removes the FAILED for test 22 on Win32.
Steve Hay [Sun, 11 Oct 2009 23:09:51 +0000]
Fix failed() calls in t/comp
Craig A. Berry [Sun, 11 Oct 2009 22:56:23 +0000]
Try for smarter (more portable) skip criteria in Encode's piconv.t.
We already differ from upstream, which skips when PERL_CORE is set.
Since we apparently think we can run it in core now, it seems like
we ought to sort out the proper skip criteria before sending it
upstream.
Craig A. Berry [Sun, 11 Oct 2009 22:45:02 +0000]
No longer need to skip t/comp/opsubs.t #22 on VMS.
And it's a good thing since the skip infrastructure was pulled
out from under us with
6b077bebea000af2a5477d50d7604bab33ee75c2.
The Win32 TODO probably still doesn't work and will need to be
implemented locally without the require of test.pl.
Nicholas Clark [Sun, 11 Oct 2009 17:56:01 +0000]
porting/manifest.t now tests that all files listed in MANIFEST are present.
Nicholas Clark [Sun, 11 Oct 2009 16:28:02 +0000]
Default to running tests in cpan/ with relative paths for perl and @INC
Explicitly turn paths absolute for the 33 extensions in cpan/ that fail tests
with relative paths.
Nicholas Clark [Sun, 11 Oct 2009 14:12:15 +0000]
Run tests in ext/ and dist/ with relative paths for perl and @INC.
Also, as only tests in cpan/ are using %no_abs and %temp_no_core, only consult
these look-up hashes for tests in cpan/
Nicholas Clark [Sun, 11 Oct 2009 14:11:06 +0000]
Swap the chdir and the call to perl_lib(), to work with relative paths in @INC.
Nicholas Clark [Sun, 11 Oct 2009 14:05:58 +0000]
MakeMaker::Test::Utils::perl_lib now copes with relative paths for core testing.
In the core, @INC already contains the moral equivalent of blib/lib. However,
it's a relative path (by default), so make it absolute. It's easier to KISS if
this is done *before* any change of directory, so document this, and change the
non-core case to add the absolute path of 'blib/lib' to @INC, rather than the
absolute path of '../blib/lib'.
Nicholas Clark [Sun, 11 Oct 2009 13:09:02 +0000]
Remove core @INC setting and chdir boilerplate from Thread::Semaphore's tests.
Nicholas Clark [Sun, 11 Oct 2009 13:07:03 +0000]
Remove core @INC setting and chdir boilerplate from Thread::Queue's tests.
Nicholas Clark [Sun, 11 Oct 2009 13:06:22 +0000]
Remove core @INC setting and chdir boilerplate from SelfLoader's tests.
Nicholas Clark [Sun, 11 Oct 2009 13:05:13 +0000]
Remove core @INC setting boilerplate from lib's test.
Nicholas Clark [Sun, 11 Oct 2009 10:05:30 +0000]
Correct the location of the DISTRIBUTION tarball for Parse::CPAN::Meta
Possibly we shouldn't store this, if it can be determined unambiguously from the
PAUSE index files.
Nicholas Clark [Sun, 11 Oct 2009 09:19:03 +0000]
Exclude CGI/t/fast.t from the core, as it relies on FCGI, which is not in core.
Whilst the test is designed to skip if FCGI is not present, that logic is now
broken, and even if it were fixed, the as-implemented skip count is now wrong.
We're never going to run this test, as core tests run with just the core's
libraries in @INC, so I see no reason to ship it.
Nicholas Clark [Sun, 11 Oct 2009 07:58:13 +0000]
Correct the name of subtest.t, added in
2c4d5b9bac4cc261
There isn't actually an "all files in MANIFEST are present" test in t/porting.
Right now we rely on ./Configure on *nix doing it, which doesn't work if you
aren't on *nix, or are, but don't re-run Configure. Maybe we should add one.
Nicholas Clark [Sun, 11 Oct 2009 07:41:53 +0000]
Correct the listed names of the two files uudecoded in commit
7b742a041dc3ba8c
It removed the '.uu' from the end of the name, but didn't change the preceding
'_' to '.'. They'd been changed from '.' to avoid having multiple '.'s in the
name, as some VMS filesystems don't like multiple '.'s. It's not easy being
green.
Josh ben Jore [Sun, 11 Oct 2009 05:36:52 +0000]
Darwin's groups can have spaces in them
Steve Hay [Sun, 11 Oct 2009 02:11:15 +0000]
Fix Test-Simple's new 00compile.t test in core
It was failing even with its MANIFEST file now in place because that
MANIFEST listed an IO::Scalar file that is EXCLUDED from core.
Steve Hay [Sat, 10 Oct 2009 13:46:56 +0000]
Add Test-Simple's MANIFEST file
It is needed by the new 00compile.t test script. We don't normally include
CPAN distributions' MANIFEST files in core, but there is an existing
precedent for this in ExtUtils-MakeMaker.
Steve Hay [Sat, 10 Oct 2009 13:40:26 +0000]
Remove test boilerpates from ExtUtils-CBuilder tests
I was obviously wrong to restore in
1051afdbc0a9e60b8f65ed22fefafd8ecf771171
what Nicholas had removed in
4d984b75f1859fc27f556ee42f31a04df42a2004 because
the tests don't work in core with the boilerplates in place.
Don't undo that which has been done by Nicholas!
Steve Hay [Sat, 10 Oct 2009 13:38:16 +0000]
Correct MANIFEST, broken by Test-Simple upgrade
Steve Hay [Sat, 10 Oct 2009 12:11:04 +0000]
Upgrade to Test-Simple-0.94
Steve Hay [Sat, 10 Oct 2009 11:57:52 +0000]
Upgrade to Storable-2.21
Retained local changes to *.t (mostly test boilerplates) since blead
is canonical for this distribution
Steve Hay [Sat, 10 Oct 2009 11:47:30 +0000]
Upgrade to Pod-Simple-3.08
Steve Hay [Sat, 10 Oct 2009 11:43:02 +0000]
Sync Pod-Simple test scripts with CPAN versions
Retained local changes in:
corpus.t, reinit.t, search20.t, search22.t, search50.t
because they look more significant
Steve Hay [Sat, 10 Oct 2009 11:31:12 +0000]
Upgrade to Parse-CPAN-Meta-1.40
Steve Hay [Sat, 10 Oct 2009 11:24:52 +0000]
Upgrade to ExtUtils-Manifest-1.57
Steve Hay [Sat, 10 Oct 2009 11:22:44 +0000]
Upgrade to ExtUtils-CBuilder-0.260301
Steve Hay [Sat, 10 Oct 2009 11:20:43 +0000]
Sync ExtUtils-CBuilder test scripts with CPAN versions
This restores the test boilerplates that were removed when ExtUtils-CBuilder
was moved from lib/ to ext/ (
4d984b75f1859fc27f556ee42f31a04df42a2004)
Steve Hay [Sat, 10 Oct 2009 11:05:09 +0000]
Upgrade to CGI.pm-3.48
Steve Hay [Sat, 10 Oct 2009 10:55:42 +0000]
Replace UU encoded files in CGI with their binary originals
There is no longer any need to avoid having binary files in the perl
distribution, and these files are not UU encoded in CGI on CPAN.
Steve Hay [Sat, 10 Oct 2009 10:50:41 +0000]
Move CGI's Changes file to its proper place
Jesse Vincent [Sat, 10 Oct 2009 20:48:38 +0000]
Applied a patch from David Fifield to fix an error message in perlipc.pod
Subject: Comment doesn't match code in perlipc.pod.
Date: Thu, 30 Jul 2009 07:31:48 -0600
To: perlbug@perl.org
From: David Fifield <david@bamsoftware.com>
Jesse Vincent [Sat, 10 Oct 2009 20:15:35 +0000]
Added a TODO test for RT#69332 - PerlIO returns undef instead of 0 on EOL
Simon 'corecode' Schubert [Sat, 10 Oct 2009 16:55:26 +0000]
[perl #69686] Fix DragonFly thread lib selection
Hey,
attached patch fixes the thread library detection in DragonFly. All
versions of DragonFly can be used threaded with -pthread. libc_r must
not be used directly.
Please let me know if you have any questions.
Thanks,
simon
Nicholas Clark [Fri, 9 Oct 2009 20:12:16 +0000]
Make _quote_args() return a result, rather than passing it a scalar reference.
This was the only use of scalar references in test.pl, and the only use of \
to create a reference to a scalar. We are now more robust in the face of parser
or runtime bugs.
Nicholas Clark [Fri, 9 Oct 2009 20:03:17 +0000]
Avoid using ++, op= and anon hash constructors in the testing code.
{} could be misparsed, ++ has a lot of internal implementation "magic" that we
don't need, but don't want to trip us up if it isn't working, and op= isn't
necessary when we already rely on the more general $a = $b op $c working.
Nicholas Clark [Fri, 9 Oct 2009 19:21:35 +0000]
Refactoring tests to use test.pl is a TODO.
Nicholas Clark [Fri, 9 Oct 2009 18:24:56 +0000]
Add a test for the bootstrap rules for tests in t/
Nicholas Clark [Fri, 9 Oct 2009 17:18:52 +0000]
Don't use require in comp/fold.t, as require isn't tested yet.
Emit TAP directly.
Nicholas Clark [Fri, 9 Oct 2009 17:16:35 +0000]
Give names to all tests in t/comp/fold.t
Nicholas Clark [Fri, 9 Oct 2009 16:57:12 +0000]
Move the require './test.pl' to the end of t/comp/hints.t
Ideally tests in t/comp wouldn't use require, as require isn't tested yet, but
this test really needs runperl(), and really wants to live in t/comp/hints.t,
so place it at the end, so that any catestrophic failure only fails the last
test. We don't use any other functionality of t/test.pl
This test uses hard-coded test numbers, but I'm not convinced that it would be
correct to re-write it to use an automatically incrementing counter, as that
wouldn't fail in an obvious fashion if some compile-time blocks ran out of
order. What we have *will* fail in an informative fashion if compile time
blocks do not run correctly.
Nicholas Clark [Fri, 9 Oct 2009 16:35:59 +0000]
Don't use require in comp/multiline.t, as require isn't tested yet.
Emit TAP directly.
Nicholas Clark [Fri, 9 Oct 2009 14:07:22 +0000]
Don't use require in comp/opsubs.t, as require isn't tested yet.
Emit TAP directly.
Nicholas Clark [Fri, 9 Oct 2009 12:17:29 +0000]
Don't use require in comp/our.t, as require isn't tested yet.
Emit TAP directly.
Nicholas Clark [Fri, 9 Oct 2009 12:13:55 +0000]
Don't use require in comp/parser.t, as require isn't tested yet.
Emit TAP directly.
Nicholas Clark [Fri, 9 Oct 2009 12:06:07 +0000]
Don't use require in comp/retainedlines.t, as require isn't tested yet.
Emit TAP directly.
Nicholas Clark [Fri, 9 Oct 2009 11:54:30 +0000]
Avoid relying on prototypes working for tests to pass. They aren't tested yet.
Nicholas Clark [Fri, 9 Oct 2009 11:48:43 +0000]
Don't use require in comp/uproto.t, as require isn't tested yet.
Emit TAP directly.
Nicholas Clark [Fri, 9 Oct 2009 11:44:10 +0000]
In opt(), use is(..., undef) rather than ok(!defined ...)
Nicholas Clark [Fri, 9 Oct 2009 10:54:29 +0000]
Move the test for require 5.11.0 not loading strictures to require.t from use.t
Nicholas Clark [Fri, 9 Oct 2009 10:49:05 +0000]
Don't use require in comp/utf.t, as require isn't tested yet.
So emit TAP directly, rather than utilising test.pl. Like test.pl, avoid using
++, as it has complexity, and that complexity isn't tested yet.
Jesse Vincent [Fri, 9 Oct 2009 16:38:48 +0000]
Record that Ask and Steve Hay have volunteered to be releng victims
Steve Hay [Fri, 9 Oct 2009 11:44:11 +0000]
Add missing IO-Compress test file
The original IO-Compress patch:
http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2009-03/msg00293.html
omitted the file (but strangely included the entry "t/cz-03zlib-v1.t*" [sic]
in the MANIFEST file that it mistakenly included).
Also note in Maintainers.pl that IO-Compress is now in cpan/ not dist/,
and mark UPSTREAM as 'cpan'.
Nicholas Clark [Fri, 9 Oct 2009 09:03:41 +0000]
Use require.t's bytes_to_utf() in place of PerlIO layers in utf.t
This avoids it needing to load Config, and means it can by run with miniperl,
and PerlIO-disabled perls.
Nicholas Clark [Fri, 9 Oct 2009 08:52:43 +0000]
Print the encoding name as part of the test, rather than on a separate line.
Nicholas Clark [Fri, 9 Oct 2009 08:48:57 +0000]
Refactor bytes_to_utf16() into a more generic routine that also handles UTF-8
Remove the direct code for testing UTF-8, calling bytes_to_utf() from a loop for
all 3 tested encodings.
Nicholas Clark [Fri, 9 Oct 2009 08:18:37 +0000]
Replace longhand invocations of test() with 3 nested loops.
Nicholas Clark [Fri, 9 Oct 2009 08:16:15 +0000]
Test utf8 with BOMs, like we already test utf16be and utf16le.
Nicholas Clark [Thu, 8 Oct 2009 19:58:17 +0000]
Move tests for use for "new style version numbers" to use.t from require.t
Jesse Vincent [Fri, 9 Oct 2009 04:13:06 +0000]
Slightly clean up the release manager guide by specifying an "End here for SNAPSHOT" per DAPM
Jesse Vincent [Fri, 9 Oct 2009 04:03:37 +0000]
Document that blead is released by a member of the sucker-of-the-month club.
Jesse Vincent [Fri, 9 Oct 2009 03:57:46 +0000]
Update the release manager guide to clean up the tagging instructions
dapm++ pointed out that I'd duplicated the "create a tag" instructions
while doing 5.11.0. Because of dmq's new tag-based releng
infrastructure, the tag step needed to move earlier.
Craig A. Berry [Thu, 8 Oct 2009 23:49:04 +0000]
MakeMaker testing in core runs from one level deeper than before.
So when we set up a dummy root for the tests on VMS, we need to
make it two levels above current rather than one in order to
prevent a lot of test failures that say "PERL_CORE is set but
I can't find your PERL_SRC!"
Nicholas Clark [Thu, 8 Oct 2009 19:05:34 +0000]
Move &do_require to the top of require.t, so that @a is the first lexical.
The relevant line, with comment, is:
my @a; # magic guard for scope violations (must be first lexical in file)
Nicholas Clark [Thu, 8 Oct 2009 18:30:08 +0000]
Move the test for RT #49472 to op/attrs.t from comp/require.t
Rafael Garcia-Suarez [Thu, 8 Oct 2009 18:11:41 +0000]
Disable strictures while setting $VERSION in a "package" statement
Rafael Garcia-Suarez [Thu, 8 Oct 2009 13:13:18 +0000]
Merge branch 'feature/package-name-version' into blead
Rafael Garcia-Suarez [Thu, 8 Oct 2009 13:12:11 +0000]
Documentation typo
Rafael Garcia-Suarez [Thu, 8 Oct 2009 12:56:56 +0000]
Simplify code that sets $VERSION on the "package" line
Nicholas Clark [Thu, 8 Oct 2009 09:10:30 +0000]
Avoid using bytes in require.t, as pack "C0U" does what we need.
Nicholas Clark [Thu, 8 Oct 2009 08:51:36 +0000]
Avoid using the strict pragma in retainedlines.t - use may not work yet.
Nicholas Clark [Thu, 8 Oct 2009 08:47:00 +0000]
Avoid using the strict pragma in colon.t - use may not work yet.
Nicholas Clark [Thu, 8 Oct 2009 08:26:01 +0000]
Avoid using pragmata in fold.t - use may not work yet.
Nicholas Clark [Thu, 8 Oct 2009 08:21:42 +0000]
Avoid using pragmata in opsubs.t - use may not work yet.
Nicholas Clark [Thu, 8 Oct 2009 08:18:31 +0000]
Avoid using the warnings pragma in proto.t - use may not work yet.