This Week on perl5-porters - 26 January-1 February 2009

Yves Orton reminded Karl Williamson that the filenames produced by the Unicode build phase need to behave correctly in the light of case-insensitive filesystems. Karl explained that mktables goes as far as ensuring compatibility with 8.3 filesystems.

Deprecate UNIVERSAL->import

Ricardo Signes introduced a lexical "this is deprecated" warning for people who write UNIVERSAL->import, since they're not supposed to anyway. Surprisingly, the test suite didn't complain.

5.10.0 Source Tree Update Problem

Dave Mitchell needs to merge the git config tweaks in blead back into 5.10 to allow testers to smoke anew. Jerry D. Hedden pointed to a patch that needs to be applied to fix the patchnum situation.

Storable::nstore_fd + '>:raw' pragma

David Williams reported a 20x slowdown when writing out a Storable data structure. Eric Brine and Bo Lindbergh suggested that this was more a problem of documentation than anything else.

Make t/win32/system.t use the more common compiler check

Max Maischein tweaked the Win32 test suite to follow the existing best practices for identifying the compiler used. Andy Dougherty pointed out that the current best practices were sub-optimal and there was a better way, that Win32 would do well to adhere to, even if Unix-like platforms don't right now.

In the final analysis, Andy wondered why tests even cared which compiler was used to build perl.

Git commits and branches

Jerry D. Hedden was puzzled by a commit reporting itself as apparently belonging to a 5.8 change, and wondered why it didn't in fact report itself as a 5.10 change. Rafa�l Garcia-Suarez explained why it was so, and Sam Vilain explained what it would take in order straighten out the history

Understanding merge history

And on the subject of git, Dave Mitchell professed to being a bit lost in the brave new world of git, and was looking for clues to help him figure out what (and how) to cherry-pick commits from blead for 5.10.

The problem appears to be sufficiently hairy that, while there is little doubt that git can do it, the question is how to ask for it.

perldelta omission for 5.8.9?

Ricardo Signes was surprised the qr/.../m improvement in blead and 5.10 made it back in time for the 5.8.9 release, and that it didn't rate a mention in the perldelta file, since it is a behavioural change.

PerlIO::via open fails for certain names

Scott Lanning want to use IO layers and ran into peculiar problems that were difficult to diagnose. After some discussion, Rafa�l thought he had a fix for the matter.

CPAN.pm 1.93_03

[RFC] Bringing Perl more into line with Unicode 5.1

[blead-473-g3b333d0] vms stat patches (take 2)

[Encode] 2.28 released!

--[Encode] 2.29 released! -- to make PAUSE happy

CPAN.pm 1.93_03

TODO of the week

Patches of Interest

-----Re: [PATCH] show -E in error message when called with -E

---Re: [PATCH] Fix mktables time stamp test

-Re: [PATCH] Code refs aren't lockable

-Re: [PATCH: perl@8482] partial support for dynaloading on OS/390

Re: [PATCH] Re: [PATCH] File::Copy & permission bits.

[PATCH] Re: Fix memory leak I introduced, forgot attachment

[PATCH] setpgrp() should extend the stack before modifying it

[PATCH] pare down UNIVERSAL docs in perlobj

[PATCH] Silence verbosity in lib/File/Copy.t during 'make test' in core

[PATCH] Makefile in lib/unicore didn't clean sub-sub directories

[PATCH] Use Unicode 5.1 files consistently

[PATCH] DB_File 1.818

[PATCH] Fix array out-of-bounds error in utfebcdic.h

Smoke [5.11.0] 2009-01-27.07:38:08 FAIL(F) netbsd 5.0_BETA (i386/2 cpu)

Smoke [5.10.0] 2009-01-06.00:10:05 FAIL(M) MSWin32 WinXP/.Net SP3 (x86/2 cpu)

Smoke [5.11.0] 2009-02-31.21:38:08 FAIL(F) openbsd 4.4 (i386/1 cpu)

Smoke [5.11.0] 2009-01-31.21:38:08 FAIL(F) openbsd 4.4 (i386/1 cpu)

New and old bugs from RT

Re: Including XSUB.h before perl.h gives nasty error (#57176)

'Data::Dumper' reports double value as string. (#62694)

pack/unpack error (#62696)

large speed difference in simple regex between pcre and GNU regex (#62718)

Perl_my_atof does not "correctly round" (#62746)

ugly English docs (#62848)

printf does not print a warning when a lone %s conversion is used (#62874)

Perl5 Bug Summary

This is the BBC

New Core Modules

This Week on perl5-porters - d-d Month 2009

Topics of Interest

--Re: perldelta5101

-----Where is Cross.pm? (was Re: merging make_ext and make_ext_cross)

-----Re: Need advice on gotchas on upgrading unicode db to 5.1

-Re: [blead-358-g3319632] vms glob patches

----Re: a plan for ext/

-Re: a plan for pod/

Works On My Machine: lib/Tie/RefHash/threaded.t passes on Win32/MinGWgcc

Re: patch@blead vms fileify_dirspec refactor / Unix mode fixes

Unbreak build on Windows

doc patch for PerlIO

What should /\p{CanonicalCombiningClass:AL}/ match?

Fix memory leak I introduced

deprecate UNIVERSAL->import

5.10.0 Source Tree Update Problem

Storable::nstore_fd + '>:raw' pragma

int localtime_r(const time_t*, struct tm*);

Make t/win32/system.t use the more common compiler check

[blead-442-gc6d234b] VMS stat fixes.

Git commits and branches

understanding merge history

perldelta omission for 5.8.9?

Re: perldelta omission for 5.8.9?

[blead-435-ga979ce9] vmspath without thread context.

Re: do_exec() is a mathom, so call its replacement directly

a condition that is always true?

Re: a condition that is always true?

RE: a condition that is always true?

PerlIO::via open fails for certain names

[blead-473-g3b333d0] vms stat patches (take 2)

Re: merging make_ext and make_ext_cross

[RFC] Bringing Perl more into line with Unicode 5.1

[blead-473-g3b333d0] vms stat patches (take 2)

Patches of Interest

-----Re: [PATCH] show -E in error message when called with -E

---Re: [PATCH] Fix mktables time stamp test

-Re: [PATCH] Code refs aren't lockable

-Re: [PATCH: perl@8482] partial support for dynaloading on OS/390

Re: [PATCH] Re: [PATCH] File::Copy & permission bits.

[PATCH] setpgrp() should extend the stack before modifying it

[PATCH] pare down UNIVERSAL docs in perlobj

[PATCH] Silence verbosity in lib/File/Copy.t during 'make test' in core

[PATCH] Makefile in lib/unicore didn't clean sub-sub directories

[PATCH] Use Unicode 5.1 files consistently

[PATCH] DB_File 1.818

[PATCH] Fix array out-of-bounds error in utfebcdic.h

Watching the Smoke Signals

Smoke [5.11.0] 2009-01-27.07:38:08 FAIL(F) netbsd 5.0_BETA (i386/2 cpu)

Smoke [5.10.0] 2009-01-06.00:10:05 FAIL(M) MSWin32 WinXP/.Net SP3 (x86/2 cpu)

Smoke [5.11.0] 2009-02-31.21:38:08 FAIL(F) openbsd 4.4 (i386/1 cpu)

Smoke [5.11.0] 2009-01-31.21:38:08 FAIL(F) openbsd 4.4 (i386/1 cpu)

New and old bugs from RT

Including XSUB.h before perl.h gives nasty error (#57176)

Data::Dumper reports double value as string (#62694)

pack/unpack error (#62696)

Perl_my_atof does not "correctly round" (#62746)

Chris Hall pointed out some inconsistencies in Perl's implementation of text-to-native-floating-point conversions. One minor hassle that Nicholas Clark remembers about the matter is that it is impossible to have an implementation compatible with both C89 and C99 since they are mutually inconsistent.

ugly English docs (#62848)

printf does not print a warning when a lone %s conversion is used (#62874)

Perl5 Bug Summary

New Core Modules

In Brief

Dave Mitchell is starting the arduous task of producing perldelta5101, the changes in perl 5.10.1.

Nicholas Clark would like to know where is Cross.pm, in the context of merging make_ext and make_ext_cross.

Max Maischein has probably succeeded in getting lib/Tie/RefHash/threaded.t to pass on Win32/MinGWgcc.

Max also unbroke the build on Windows that had suffered at the hands of to a refactoring of make_patchnum.pl.

That Karl Williamson doesn know what /\p{CanonicalCombiningClass:AL}/ should match probably points to an error in mktables.

Last week's summary

About this summary

If you found this summary useful, please consider contributing to the Perl Foundation or attending a YAPC to help support the development of Perl.