This Fortnight on perl5-porters - 21 June-5 July 2008

There is a problem with format and write and forward declarations, due to a jumbo closure patch from Dave Mitchell in 2003. Apparently, due to the rewrite of the closure and lexical code in 5.10, what used to be simply a wrong result now become a core dump.

=cut toggles POD/code compilation rather than forcing code (#24483)

Ren�e wrote some code to solve the fact that a =cut following code will case the parser to flip into POD mode, rather than generating a warning. Abigail and Rafa�l both admitted to having relied in the existing behaviour, and that it was more likely a task for Perl::Critic or Pod::Checker to solve.

for any perl thing, perldoc -f should cough up some tip (#27886)

Ren�e also set about making perldoc -f return something useful for things like operators (such as tr), and welcomed people's suggestions.

perldoc lib doesn't describe actual behaviour of use lib (#34339)

Ren�e added some documentation to address the concern. Steffen M�ller updated the CPAN dual-lifed version, and hinted that blead's version needs to be bumped up a couple of notches.

open(FILEHANDLE, "<&=$fd") + dupping in memory files (#50756)

Bram looked at this report, and suggested that part of the problem was that the original submitter made a mistake, and the remaining problem was more a documentation issue than anything else.

File::Glob fails to set error (#51410)

Nicholas Clark filed this report a couple of months ago, in an attempt to nail down an intermittent failure with File::Glob. Bram wondered whether running the test a few more times might trigger a smoke failure on some other platform.

s/// on tied hashes (#51514)

Bram saw that the failure that cropped up in 5.10 is now fixed in blead, at least on his machine, and so he asked the original submitter to confirm that it was indeed the case for them.

PERL_NO_SHORT_NAMES incompatible with perl.h (#51762)

eval'ing a certain format string segfaults perl (#55668)

Nicholas Clark reported that the bug was still with us, and pointed to the exact section of code that was no doubt responsible for the problem.

return return (#56150)

At the end of this thread from the week before, Reini Urban achieved the impressive feat of getting the tokeniser and the peep-hole optimiser to emit a warning when a doubled-up return was used. Having done that, he wasn't sure what to do with it; certainly no-one had the courage to apply the patch.

(((??{1 + $^N}))) behaves differently in 5.10.0 to blead (#56194)

Bram noted that reverting change #32857 was sufficient to restore the previous behaviour.

Unexpected UTF-8 flag in ascii strings (#56336)

This bug report fell out of a thread on Perlmonks, but Rafa�l Garcia-Suarez was disinclined to act upon it, considering it as a minor implementation detail.

encode() with FB_CROAK may destroy its argument (#56340)

Alex Kapranoff found a problem in Encode::encode when playing with UTF-7 strings, and suggested a fix.

Alex also found that the UTF-7 codec in Encode uses m/\G/g without resetting pos() (bug #56342).

glob() with spaces - documentation unclear (#56348)

Ed Avis tripped over the behaviour of spaces in glob parameters, but Paul Fenwick was quick to point out that it's a very desirable feature. He tightened up the documentation in File::Glob, which Rafa�l applied.

odd segmentation fault with overload.pm in perl 5.8.8 (#56376)

Paul Miller reported a bug in overload.pm, only to discover that the porters had already encountered the error in an earlier incarnation and it was already fixed in blead.

Odd behavior of given {when (... or ...) ...} (#56416)

"Polynomial" posted a question over at Perlmonks regarding some mysteriously inconsistent results for smart matching. Since none of the answers solved the matter conclusively, a bug report was filed.

Perl 5.10 breaks \N{} in regex interpolated inside regex (with charnames) (#56444)

Chris Pirazzi found a problem in 5.10 that Andreas Knig traced to a patch from Yves Orton back in September 2006. Rafa�l discovered that a judicious use of eval could sweep the problem under the carpet, thus leading him to conclude that there was some problem involving the order of loading afoot.

%+ still untaints data when use re qw(taint) is in scope (#56490)

perl5.10.0: caching environment causes panic in some modules (esp. Term::ReadLine::Gnu) (#56500)

Empty list elements in list context (#56524)

m/a{1,0}/ compiles but doesn't match a literal string (#56526)

<variable> used only once: possible typo (#56530)

pad_alloc and fold_constants are not members of public API (#56536)

5.10.0 crash on serialised v-strings (#56606)

Math::BigInt gives completely when Math::BigInt::GMP is used (#56608)

Perl 5.10.0 double-free bug with regular expressions (#56610)

Perl5 Bug Summary

New Core Modules

In Brief

Ren�e B�cker wondered what to do about really old tickets in RT that appear to contain wrong information, and the issuer never replied to a request for more information in over four years. Bram suggested that such tickets should be rejected, on the grounds that if it were a serious problem, someone would have filed a better bug report in the meantime.

Juerd Waalboer wrapped up the Perl 5.8 and perl 5.10 differences on UTF/Pack things thread with a useful review for dealing with legacy code that deals poorly with UTF-8 strings.

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.