This Week on perl5-porters (19-25 January 2004)

This Week on perl5-porters (19-25 January 2004)

Another summary this week, not really different from the previous ones, as it contains the usual bug reports and other discussions. The bugs, however, are new. Read below.

On the order of arguments to Exporter

Mark-Jason Dominus reports that you can't write

    use Fcntl "O_RDONLY", ":flock";

because Exporter will think that :flock is the name of a symbol to be imported, not of a group of symbols. (bug #25160.) H. Merijn Brand committed a patch to change this.

The dor patch to 5.8.3

H. Merijn Brand uploaded the defined-or patch against 5.8.3 to CPAN (which adds to it the // and err operators that exist in bleadperl). However, it contained a minor error, causing the output of perl -V not to include the proper Locally applied patches section. As PAUSE doesn't let people upload the same file twice, Merijn deleted the first patch and uploaded a new one under the name dor-5.8.3a.diff, which can be found in his CPAN directory (HMBRAND).

In Brief

Leon Brocard released a snapshot of perl 5.005_04 :

Some progress was made on the Storable module. Stas Bekman sent a patch to make it thread-safe. Moreover, Sam Vilain fixed a case of segmentation fault (bug #25145).

Stas wants the strategies used to handle memory allocations on Windows to be improved, esp. where ithreads are involved. Jan Dubois is working on it, but issues are complex.

Alan Burlison points out an error in the documentation of threads pertaining to the interaction of threading with forking. Also, he notes that an explicit unlock() would be more flexible than the implicit unlock-at-end-of-scope that happens currently.

Steve Hay proposed a patch to add a camel icon to perl.exe on Windows.

David Jantzen reports (bug #25237) that perl leaks memory when recursively traversing circular datastructures. Dave Mitchell explains that he managed somehow to create a closure which closes over itself. So leakage is to be expected.

Bugs fixed

Gisle Aas reported that the simple statement

    print if foo

doesn't die under strictures (bug #25147). This was fixed by Dave Mitchell.

The constructs

    sub foo : unique {}
    my $x : unique;

are now disallowed in bleadperl (until someone comes up with a sensible signification for them.)

Detached threads now work on Windows, thanks to Jan Dubois. They used to crash on interpreter exit.

About this summary

This summary was written by Rafael Garcia-Suarez. Weekly summaries are published on and posted on a mailing list, which subscription address is Corrections and comments are welcome.