The rhythm of maintenance releases is now well established, and this week saw the release of perl 5.8.4, as expected. Meanwhile, the usual stream of bugs and patches continued.
Perl 5.8.4 was released. The new setuidperl executable which was present in RC2 was finally withdrawn; the new plan is to have suidperl installed as a hard link to perl, and sperl5.8.4 being the actual set-uid executable. (See the latest perldelta for the details.)
Yves Orton tells the crowd about the newly released command-line compiler by Microsoft, available as a free download. Follows some remarks on how to build perl on Windows with it, and what are the odds and ends of the differents versions of Microsoft's compilers.
Nicholas Clark ran a tool to find duplicated code on the perl source code; and it seems that there's some room for refactorization. As he writes, quite a bit is duplication between code in the win32, wince and netware subdirectories.
Dave Mitchell commited to bleadperl his patch to enhance the Use of uninitialized value warning by adding the name of the variable or array/hash subscript, if any, which was uninitiliazed. Nicholas Clark wonders if it's applicable to the perl 5.8.x branch, but Dave points out that it's likely to cause spurious test failures in various CPAN modules. Notably, he had to patch several core tests corresponding to CPAN modules to make them cope with the new form of the warning.
Marcus Holland-Moritz committed a patch to add the support in pack() templates for the
> modifiers which were discussed in a previous summary. He has some questions on some oddities of the implementation, but nobody commented. (Anybody has a Cray handy?)
Jeff Pinyan proposed a patch to bleadperl to enhance perl's Unicode tables, notably for a better compliance.
Stas Bekman supplied a patch to log the perl interpreter address in internal warnings emitted by a multithreaded perl.
Jay Hannah is surprised by the fact that sort() in scalar context doesn't return any useful value (more precisely when it's used as the last thing in a function that gets called in scalar context). Randal Schwartz points out that there is no such thing as a list value in scalar context, and gives a list of builtins commonly used in a list context, and what special value they return in a scalar context.
David Manura found a bug related to the pos() corresponding to a match done on a reference to a substr() lvalue. Probably some magic that is not cleaned as it should, according to Yitzchak Scott-Thoennes.
Mark Jason-Dominus reports that File::Find doesn't detect dangling symlinks when the option
follow_fast is used (bug #28929).
This summary was written, in the hope you enjoy it, by Rafael Garcia-Suarez. Weekly summaries are published on http://use.perl.org/ and posted on a mailing list, which subscription address is email@example.com. Comments and corrections welcome.