This Week on perl5-porters (30th December 2002 / 5th January 2003)

dev.perl.org
 

This Week on perl5-porters (30th December 2002 / 5th January 2003)

What are the perl5-porters' new year resolutions ? You won't learn it by reading this week's highly technical summary. But that's not the point, is it ?

Copy constructor contract

Rick Delaney reports (bug #19582) that a copy constructor that doesn't return a blessed reference may produce segfaults when the resulting reference is used. As a solution, he proposes to forbid copy constructors to return non-objects. (Currently, copy constructors can't return non-references, but can return unblessed references.)

  The bug : http://archive.develooper.com/perl5-porters@perl.org/msg90340.html
  The fix : http://archive.develooper.com/perl5-porters@perl.org/msg90321.html

Simple segfault

Some good bug hunting was done for bug #19566, a simple case of segfault, caused by this simple snippet of code (there are many variants) :

    for (\2) { $_ = <FH> }

Oddly enough, this bug seems to exist since a long time (5.003_01), but haven't been fixed.

    http://archive.develooper.com/perl5-porters@perl.org/msg90334.html

Parens in pack()

Wolfgang Laun asked a question about the semantics of parentheses in pack() templates. Nicholas Clark confessed that '(' in pack makes his head hurt, and frankly, I'm with him on this point. See for yourself :

    http://archive.develooper.com/perl5-porters@perl.org/msg90455.html

Link black magic

Chip Turner initiated a thread on shared libperl builds, which I don't (frankly) want to summarize here. If you're interested in the issues in packaging, distributing and upgrading binaries of perl, follow the link :

    http://archive.develooper.com/perl5-porters@perl.org/msg90384.html

In brief

Gisle Aas provided updates for his modules MIME::Base64 and Digest::MD5.

Hugo released a snapshot of bleadperl (the 5.9.0-to be), shortly followed by Jarkko releasing a snapshot of maintperl (the 5.8.1-to be). Jarkko released another snapshot at the end of the week before his planned connectivity problems.

Nicholas Clark proposed a patch to implement copy on write for the $& and $<DIGIT> variables. As with other copy on write implementations, it's available only if perl is configured with the -DPERL_COPY_ON_WRITE flag. However, this patch doesn't seem to be a significant speedup (or slowdown).

On the performance chapter, Jarkko also sent a patch to speed up Unicode operations.

Jarkko thrown at us an interesting link : Someone might want to try Splint at Perl 5 (or Parrot, for that matter) sources. (see http://www.splint.org/).

About this summary

This summary brought to you by Rafael Garcia-Suarez and his hurting brain. (OK, I was obviously tired this week, I blame this bad cold I got.) Summaries are available on http://use.perl.org/ and/or via a mailing list, which subscription address is perl5-summary-subscribe@perl.org. Comments and corrections are welcome.