Craig A. Berry [Mon, 23 Feb 2009 00:28:13 +0000]
time_t is unsigned 32-bit on VMS; pick sGMTIME_max and sLOCALTIME_max accordingly.
Nicholas Clark [Sun, 22 Feb 2009 22:17:47 +0000]
For S_incpush(), dir is never NULL, and len is always > 0.
Nicholas Clark [Sun, 22 Feb 2009 21:27:18 +0000]
Optimise S_incpush() by avoiding repeatedly copying libdir to subdir.
Specifically, copy it once with newSVsv(), then pass libdir to
S_incpush_if_exists(), and if that creates a new SV, use newSVsv() there to
re-do the copy. Otherwise reset the length of the passed-in SV (which is
subdir), back to the length of libdir, effectively truncating it back to be
equal to libdir. This avoids repeated copying of the same bytes over the same
memory that already holds those bytes.
Nicholas Clark [Sun, 22 Feb 2009 21:11:19 +0000]
In S_incpush, move the declaration of subdir inside the only block that uses it.
Nicholas Clark [Sun, 22 Feb 2009 21:00:03 +0000]
Tidy up incpush. Re-order the macros, and re-indent some code.
Nicholas Clark [Sun, 22 Feb 2009 20:45:04 +0000]
Move unversioned directories in PERL_OTHERLIBDIRS later in @INC.
Split INCPUSH_ADD_SUB_DIRS into INCPUSH_ADD_VERSIONED_SUB_DIRS and
INCPUSH_ADD_ARCHONLY_SUB_DIRS, and tweak the flags used for the two halves of
PERL_OTHERLIBDIRS so that .../$archname and .../ come after .../$old_version
Nicholas Clark [Sun, 22 Feb 2009 20:29:32 +0000]
Nicholas Clark [Sun, 22 Feb 2009 19:27:48 +0000]
In S_init_perllib(), push the bare VENDORLIB_STEM onto @INC after its subdirs.
Nicholas Clark [Sun, 22 Feb 2009 19:27:08 +0000]
In S_init_perllib(), push the bare SITELIB_STEM onto @INC after its subdirs.
Nicholas Clark [Sun, 22 Feb 2009 19:24:49 +0000]
In S_incpush(), move push @INC, ".../$archname" if -d ".../$archname"; later.
This effects this change:
--- Start 2009-02-21 14:04:40.
000000000 +0100
+++ /dev/fd/63 2009-02-22 16:41:18.
000000000 +0100
@@ -1,14 +1,14 @@
/Users/nick/Sandpit/511v/hashbangI/5.11.0/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/hashbangI/5.11.0
-/Users/nick/Sandpit/511v/hashbangI/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/hashbangI/5.11.-2/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/hashbangI/5.11.-2
+/Users/nick/Sandpit/511v/hashbangI/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/hashbangI
/Users/nick/Sandpit/511v/cliI/5.11.0/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/cliI/5.11.0
-/Users/nick/Sandpit/511v/cliI/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/cliI/5.11.-1/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/cliI/5.11.-1
+/Users/nick/Sandpit/511v/cliI/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/cliI
/Users/nick/Sandpit/511v/perl5lib0/5.11.0/darwin-thread-multi-64int-2level
/Users/nick/Sandpit/511v/perl5lib0/5.11.0
Nicholas Clark [Sun, 22 Feb 2009 17:37:26 +0000]
In S_init_perllib(), initialise perl5lib to NULL, to keep trace flow happy.
Nicholas Clark [Sun, 22 Feb 2009 17:24:04 +0000]
In S_incpush(), replace (addsubdirs || addoldvers) with a constant variable.
Nicholas Clark [Sun, 22 Feb 2009 13:05:38 +0000]
In S_init_perllib(), s can be const char*
Nicholas Clark [Sun, 22 Feb 2009 12:55:49 +0000]
In S_init_perllib(), only call PerlEnv_getenv("PERL5LIB") once.
Nicholas Clark [Sun, 22 Feb 2009 12:34:29 +0000]
Unwind the implicit loop in S_init_perllib(), by writing the code out longhand.
Call it only once, remove the old_vers parameter, and all the related
conditional code.
Nicholas Clark [Sat, 21 Feb 2009 17:23:46 +0000]
Change the value passed into S_init_perllib() to a boolean flag.
Nicholas Clark [Sat, 21 Feb 2009 07:06:06 +0000]
NetWare has PRIVLIB_EXP as something other than a string constant. Cope with it.
Craig A. Berry [Sat, 21 Feb 2009 20:46:27 +0000]
Use "system" gmtime in time64 on VMS so C<vmsish 'time'> pragma works.
It's not really the system gmtime; it's the homegrown one in vms/vms.c
that checks hints and does shifting between UTC and local time when
requested.
Nicholas Clark [Sat, 21 Feb 2009 18:04:08 +0000]
In S_init_perllib(), correct errors in the MACOS_TRADITIONAL conditional code.
(Errors that originated in
a26c0e281cb6068a8d148933281d8186f1eb4206 and
50d61629dc7fe34f077b9f66c50287d839e06378)
Nicholas Clark [Sat, 21 Feb 2009 08:41:14 +0000]
Re-apply changes
4adc95e616bac7ee and 200cbd6aa595a074 (RT #61492)
(ExtUtils::MM_Win32 should not generate "mt" command when CRT is statically
linked)(We'd failed to send this one upstream. Bad us.)
Nicholas Clark [Sat, 21 Feb 2009 08:11:13 +0000]
Upgrade to ExtUtils::MakeMaker 6.49_01
Craig A. Berry [Sat, 21 Feb 2009 04:58:55 +0000]
Skip time64 when there is no 64-bit integer type available.
The y2038 code currently requires that a 64-bit integer type be available,
so the build falls down hard if we try to use it where there isn't one. A
better fix might be to try NV for time_t and a 32-bit integer for year.
N.B. We test for the definedness of Quad_t rather than HAS_QUAD since the
latter is explicitly undef'd in the case where 64-bit types are available
but are not being used for Perl's IV and UV.
Craig A. Berry [Fri, 20 Feb 2009 23:43:08 +0000]
Avoid VMS long filename support on older systems that don't have it.
Craig A. Berry [Fri, 20 Feb 2009 23:37:47 +0000]
Move redefinition of lstat above its first use in vms/vms.c.
Rafael Garcia-Suarez [Fri, 20 Feb 2009 23:06:22 +0000]
More smartmatch tests
This includes a TODO for overloaded objects.
Tests will need to be augmented and better organized.
Nicholas Clark [Fri, 20 Feb 2009 20:09:16 +0000]
Add a parameter to win32_get_{priv,site,vendor}lib(), to return the length,
as we already know it, and use it in S_init_perllib() to save a strlen() in
S_incpush_use_sep().
Steffen Mueller [Fri, 20 Feb 2009 18:22:57 +0000]
Text::Wrap maintained via the CPAN release
Jerry D. Hedden [Fri, 20 Feb 2009 14:45:41 +0000]
Check for thread failure in prime number example
Rafael Garcia-Suarez [Fri, 20 Feb 2009 13:45:36 +0000]
Silence a casting warning with memchr()
Alex Davies [Fri, 20 Feb 2009 08:39:29 +0000]
Wonky example in perlthrtut
Fix at least the syntax of the example.
See : Message-ID: <
F00A81D78228413FBED223C3919BE75B@Amelie>
Leon Brocard [Fri, 20 Feb 2009 08:31:40 +0000]
Merge branch 'blead' of ssh://perl5.git.perl.org/gitroot/perl into blead
Leon Brocard [Fri, 20 Feb 2009 08:30:59 +0000]
Remove the -x from the cherry pick and show how to merge the whole branch (suggested by rgs)
Rafael Garcia-Suarez [Tue, 17 Feb 2009 12:56:43 +0000]
Reorder and complete tests for smart matching with objects
Rafael Garcia-Suarez [Tue, 17 Feb 2009 07:22:44 +0000]
Fix {%hash} ~~ %hash test
This was not parsed as a smart match, because {%hash} was taken
as a block, not a an anonymous hash.
Also, avoid to construct an anonymous hash with an odd number of
elements.
Rafael Garcia-Suarez [Tue, 17 Feb 2009 06:50:16 +0000]
Better diagnostics for the ~~ test
Read from DATA line per line, so warnings are reported from the correct
line. Make test names and error reports more readable.
Leon Brocard [Fri, 20 Feb 2009 08:25:18 +0000]
Remove an extra space
Leon Brocard [Fri, 20 Feb 2009 08:22:23 +0000]
Add a section on merging from a branch via GitHub
Dan Dascalescu [Fri, 20 Feb 2009 06:48:09 +0000]
Fixed missing word in Text::Wrap POD
(cherry picked from commit
3adac458cb1c1d41af47fc66e67b49c8dec2323f)
Paul Marquess [Thu, 19 Feb 2009 09:15:01 +0000]
[PATCH] DB_File 1.819
Small change to t/db-recno.t [RT# 43288]
Message-ID: <
00ad01c99214$89f5fa50$9de1eef0$@Marquess@ntlworld.com>
Nicholas Clark [Wed, 18 Feb 2009 15:28:12 +0000]
Remove the stub nw_get_sitelib(), and undefine SITELIB_EXP for NetWare.
(nw_get_sitelib() was hard coded to return NULL, which is a slower equivalent
to simply not defining SITELIB_EXP in the first place. I suspect a cargo cult
of win32/ without actually taking the time to understand it.)
Nicholas Clark [Tue, 17 Feb 2009 23:24:32 +0000]
Pass the length of the string to S_incpush_use_sep(), where known.
Nicholas Clark [Tue, 17 Feb 2009 22:57:56 +0000]
Tests for S_incpush_use_sep() splitting on : (or platform equivalent)
Nicholas Clark [Tue, 17 Feb 2009 20:18:39 +0000]
In runenv.t, break apart running perl and testing the output into two functions.
Nicholas Clark [Tue, 17 Feb 2009 17:11:25 +0000]
Get $ENV{PERL5LIB} into a known consistent state under TEST - delete it!
(It doesn't strike me as good that t/TEST and t/harness are inconsistent -
the former runs every test with -MTestInit; the latter sets $ENV{PERL5LIB} to
'../lib'; and on the gripping hand, we have boilerplate in every test to set
@INC. No choice seems ideal.
Nicholas Clark [Tue, 17 Feb 2009 08:48:45 +0000]
In S_incpush(), avoid "possible loss of data" warnings from the C compiler.
Nicholas Clark [Mon, 16 Feb 2009 23:55:58 +0000]
Move PERL_VERSION_STRING from patchlevel.h to perl.h
(and revert
a89d24b9fe54d44395806c16ac1c19f3c678cbbb)
It's getting too complex to work around some versions of awk which don't want
to implement the regular expressions that they are documented to implement.
Nicholas Clark [Mon, 16 Feb 2009 22:47:54 +0000]
Ignore #defines in in patchlevel.h that end the line with backslash.
(Otherwise config.sh ends up with part of PERL_VERSION_STRING, and the build
chokes to a stop.)
Nicholas Clark [Mon, 16 Feb 2009 22:27:58 +0000]
As WIN32 is already using different flags to incpush_use_sep for SITELIB_EXP,
VENDORLIB_EXP and PRIVLIB_EXP, requiring conditional compilation, it doesn't
feel like extra complexity to make explict calls in S_init_perllib() to the 3
win32_*() functions they expand to, which allows a NULL pointer check.
Nicholas Clark [Mon, 16 Feb 2009 22:21:15 +0000]
By spliting PERL_ARCH_FMT into PERL_ARCH_FMT_PREFIX and SUFFIX, we can replace
two more Perl_sv_setpvf()s with sv_setsv()/sv_catpvs() pairs.
Nicholas Clark [Mon, 16 Feb 2009 22:03:35 +0000]
As PERL_ARCH_FMT_PATH is a compile time constant, can simplify one sv_setpvf()
to sv_setsv() and sv_catpvs().
Nicholas Clark [Mon, 16 Feb 2009 22:00:14 +0000]
PERL_FS_VER_FMT is only used with the current perl version, which is known at
compile time, so replace it with PERL_FS_VERSION, a compile time constant.
Nicholas Clark [Mon, 16 Feb 2009 21:42:54 +0000]
Abolish PERL_AV_SUFFIX_FMT, which was rendered obsolescent 8 years ago with
084592ab0b76f3cbd4d089afb08ccea7ba1c9dd8.
Nicholas Clark [Mon, 16 Feb 2009 21:16:21 +0000]
Replace *printf "%d.%d.%d", PERL_REVISION, PERL_VERSION, PERL_SUBVERSION
with PERL_VERSION_STRING, which can be determined at compile time, as a
constant string.
Nicholas Clark [Mon, 16 Feb 2009 21:15:17 +0000]
Update copyright years. 2008 sneaks in because of
46807d8e809cc127621bf85d9e9
David Mitchell [Mon, 16 Feb 2009 16:37:00 +0000]
add missing perforce-era entry to Changes files
Bram [Mon, 16 Feb 2009 15:07:19 +0000]
Use tempfile() in tests
Nicholas Clark [Mon, 16 Feb 2009 11:14:28 +0000]
Don't add PRIVLIB_EXP to @INC twice.
Zsban Ambrus [Mon, 16 Feb 2009 11:18:19 +0000]
[perl #63234] [DOC PATCH] fix some missing parts of IO::Handle pod
This documents the previously undocumented fcntl and ioctl methods, and
adds a see also to the IO::File module where the documentation refers
to one of its methods.
Dan Kogai [Mon, 16 Feb 2009 10:45:53 +0000]
Upgrade to Encode 2.31
Bram [Mon, 16 Feb 2009 10:22:40 +0000]
Extra regex tests
Message-ID: <
20090215185207.gsnhhqdegckws0co@horde.wizbit.be>
Message-ID: <
20090215210634.u02f15b284ogc4s4@horde.wizbit.be>
Rafael Garcia-Suarez [Mon, 16 Feb 2009 10:09:12 +0000]
Yves states that patches to ExtUtils::Install should go to blead first
Nicholas Clark [Sun, 15 Feb 2009 21:53:50 +0000]
Ensure that the pointer to S_incpush_use_sep() is never NULL.
Nicholas Clark [Sun, 15 Feb 2009 16:51:10 +0000]
In @INC, move ARCHLIB and PRIVLIB after the current version's site_perl and
vendor_perl.
Nicholas Clark [Sun, 15 Feb 2009 16:18:34 +0000]
Loop in S_init_perllib(), only calling S_incpush*() with INCPUSH_ADD_OLD_VERS
the second time (and only for those entries at had it). Implement the loop by
calling init_perllib() twice, to avoid a rats nest of re-indenting. Add a new
flag to S_incpush() INCPUSH_NOT_BASEDIR, to supress pushing the base directory
a second time on the secnod call.
With this change, re-ordering of @INC from version-orientated to prefix-
orientated is partly complete. ARCHLIB and PRIVLIB remain at their old place in
the @INC order.
Nicholas Clark [Sun, 15 Feb 2009 14:35:36 +0000]
Refactor the separator splitting loop of S_incpush() into a S_incpush_use_sep().
Add a parameter to S_incpush() to optionally pass in the length. As S_incpush()
treats the directory parameter as const char, remove some malloc()s elsewhere
that were copying data on the assumption that it was not const safe.
Nicholas Clark [Sun, 15 Feb 2009 14:01:18 +0000]
In S_incpush(), rename the parameter from dir to p, rather than copying it.
John Malmberg [Sun, 15 Feb 2009 15:25:10 +0000]
ExtUtils::Install VMS extended character set support
Preview from https://rt.cpan.org/Ticket/Display.html?id=42149
Nicholas Clark [Sun, 15 Feb 2009 12:05:55 +0000]
Refactor S_incpush() to take 1 flags parameter, instead of 5 positional booleans
(which are impossible to remember).
Nicholas Clark [Sun, 15 Feb 2009 11:27:51 +0000]
For -I, need to also unshift version and architecture libs onto @INC (RT#6665)
(
20189146be79a0596543441fa369c6bf7f85103f only added the given directory.)
Craig A. Berry [Sat, 14 Feb 2009 19:08:33 +0000]
Clear Module::Build smoke on VMS while waiting for upstream.
Except for the version bump, this is the same as:
https://rt.cpan.org/Public/Bug/Display.html?id=42724
Nicholas Clark [Sat, 14 Feb 2009 18:53:32 +0000]
Add git_version.h to private so that make distclean deletes it.
(It was already in a different target that caused make clean to delete it.)
Nicholas Clark [Sat, 14 Feb 2009 18:30:02 +0000]
Ignore the OS X GNUmakefile*s, and t/test_state.
Nicholas Clark [Sat, 14 Feb 2009 18:29:37 +0000]
Ignore lib/NDBM_File.pm
David Mitchell [Sat, 14 Feb 2009 17:26:11 +0000]
Make git-find-p4-change only find changes done under perforce.
That way, later cherry-picking that happens to copy the 'p4raw-id:'
text won't be located by mistake.
Rafael Garcia-Suarez [Sat, 14 Feb 2009 09:04:07 +0000]
Forbid to use "foreach" as an attribute
(like all other control flow statements)
Rafael Garcia-Suarez [Fri, 13 Feb 2009 23:16:20 +0000]
More tests for coderefs and smart match
Rafael Garcia-Suarez [Fri, 13 Feb 2009 22:20:11 +0000]
Remove code that has never been used in any test
Craig A. Berry [Fri, 13 Feb 2009 21:26:00 +0000]
Handle uninitialized interpreter when performing vmsish pragma checks.
Rafael Garcia-Suarez [Fri, 13 Feb 2009 21:08:06 +0000]
$object ~~ undef should always test for definedness
(even if $object hasn't overloaded the ~~ operator)
Rafael Garcia-Suarez [Fri, 13 Feb 2009 20:57:41 +0000]
Add tests for undef smart-matching
Rafael Garcia-Suarez [Fri, 13 Feb 2009 15:59:22 +0000]
Document error "Smart matching a non-overloaded object breaks encapsulation"
That error was added in
e67b97bd974194ad616acbb7813c5631aacd6be7
Rafael Garcia-Suarez [Fri, 13 Feb 2009 14:59:28 +0000]
Remove obsolete comment; document test format
Rafael Garcia-Suarez [Fri, 13 Feb 2009 15:23:28 +0000]
More tests for array-ref smart match
Rafael Garcia-Suarez [Fri, 13 Feb 2009 15:17:30 +0000]
Remove one of the two definitions of the a_const contant sub
Rafael Garcia-Suarez [Fri, 13 Feb 2009 14:47:39 +0000]
More gitignore adjustments
Ignore lib/Attribute, now that Attribute::Handlers is under ext/.
Ignore generated Makefile.PL files for Safe and Attribute-Handlers.
I preferred putting those in specific .gitignore files instead of
ignoring Makefile.PLs in the whole ext/ hierarchy, in order to
ensure a less confusing git behaviour for future generations.
Rafael Garcia-Suarez [Thu, 12 Feb 2009 21:21:44 +0000]
Remove double slash in MANIFEST
(so Porting/manicheck does not complain)
Nicholas Clark [Thu, 12 Feb 2009 21:02:24 +0000]
Move Attribute::Handlers from lib to ext. The layout in ext is the same as the
CPAN distribution, which simplifies dual life module maintenance.
Nicholas Clark [Thu, 12 Feb 2009 20:43:49 +0000]
make_ext.pl now generates a Makefile.PL if needed.
Remove ext/Safe/Makefile.PL as a proof of concept.
Craig A. Berry [Thu, 12 Feb 2009 19:22:22 +0000]
Find extensions on VMS even when they don't have a Makefile.PL.
Jerry D. Hedden [Thu, 12 Feb 2009 16:16:15 +0000]
Unigue test file names in t/lib/warnings/9enabled
Steve Hay [Thu, 12 Feb 2009 14:50:16 +0000]
Upgrade to Math-Complex-1.56
Steve Hay [Thu, 12 Feb 2009 11:16:20 +0000]
win32/ext doesn't exist any more, so no need to look there for extensions
Steve Hay [Thu, 12 Feb 2009 10:46:32 +0000]
win32/ext doesn't exist any more, so no need to look there for tests
Florian Ragwitz [Thu, 12 Feb 2009 10:42:24 +0000]
Ignore some temporary files created by ext/DB_File.
Steve Hay [Thu, 12 Feb 2009 10:17:21 +0000]
Update .gitignore with new flat extension path
Steve Hay [Thu, 12 Feb 2009 10:14:07 +0000]
Skip tests for flattened extensions that weren't built.
Commit
6ebb0601826917b6ce1b97a2bdd6577110416c64 missed t/harness,
which Win32 uses rather than t/TEST.
Also, make the new variable my().
Rafael Garcia-Suarez [Thu, 12 Feb 2009 09:20:52 +0000]
Fix MANIFEST for CPAN 1.93_51 inclusion
Andreas Koenig [Thu, 12 Feb 2009 04:44:17 +0000]
Update CPAN.pm to 1.93_51
John Malmberg [Mon, 9 Feb 2009 14:50:29 +0000]
vms fgetname wrapper.
fgetname() does not always return the correct Unix format file
specification when the decc$filename_unix_report feature is active and
is ignoring the decc$readdir_dropdot_notype setting.
So always have fgetname() return a VMS format file specification. When
decc$filename_unix_report is active, use unixify() to convert it to the
expected syntax.
This bug shows up doing rename tests on an open file that has no file
extension with decc$filename_unix_report and decc$readdir_dropdot_notype
both active.
Message-ID: <
499042B5.4030803@gmail.com>
H.Merijn Brand [Wed, 11 Feb 2009 13:52:41 +0000]
avoid confusing gcc with -
2147483648
As Robin explained, -123UL will be (should be) parsed as -(123UL), which should make all compilers happy