develooper Front page | perl.perl5.porters | Postings from October 2003

This Week on perl5-porters (13-19 October 2003)

Rafael Garcia-Suarez
October 20, 2003 13:31
This Week on perl5-porters (13-19 October 2003)
Message ID:
This Week on perl5-porters (13-19 October 2003)
  What happens in the post-5.8.1 world ? Read about the plans for the
  (nearest than you may think) 5.8.2, 5.8.1-specific problems, and other
  Perl language and implementation questions.

Binary Compatibility
  Nicholas Clark prepares himself to release perl 5.8.2, which should fix
  the binary compatibility issue raised last week between 5.8.0 and 5.8.1.
  The discussions are highly technical, since the fix should preserve
  binary compatibility and at the same time preserve the security feature
  (the hash seed randomization) which was introduced in 5.8.1 and that
  broke binary compatibility.

  The announcement of Nicholas' first snapshot can be found at

Version numbering
  With all those version number increments around, a question was raised :
  should perl's internal version number be incremented just after the
  release of the previous version, or just before the release of the new
  version ? Several arguments were raised. Usually, the tradition is to
  bump up the version number before the first release candidate.

CPAN things
  Ilya Zakharevich posted a patch containing several improvements to the module. This led to a discussion about IPC::Open3 (don't ask me

  Meanwhile, Randal Schwartz notices that the CPAN shell of perl 5.8.0
  reports AutoLoader as "to be upgraded" since 5.8.1 is out. Andreas
  Koenig agrees that the shell should be silent about core modules that
  are not (yet) dual-lived on CPAN.

getppid() and mod_perl
  Stas Bekman reported that under mod_perl, built with perl 5.8.1 on
  Linux, the perl built-in getppid() doesn't return the actual PPID of the
  httpd process, as it should. This was the consequence of a change in
  perl 5.8.1, that now caches the result of getppid(2) in an interpreter
  variable, which is not updated when apache forks. (The rationale behind
  this change is that the implementation of getppid(2) on Linux is not
  POSIX-compliant when it comes to threads, and that Perl must be POSIX
  compliant when it can.) Rafael Garcia-Suarez, responsible for the PPID
  caching, provided a patch to mod_perl to fix this, and noted that a
  similar problem potentially occurs with the $$ special variable
  (although it was already correctly handled by mod_perl.)

Dup'ing stderr
  Abigail filed bug #24204 : "older versions of Perl allowed you to dup
  STDERR for reads, but newer versions (i.e. 5.8.x) don't." This
  disallow the trick to open STDERR for reading from the keyboard when
  standard input has been redirected to read from a file.

  The discussion continues on the cleverness of Unix tricks, how and why
  they should be supported by perl, and (permitted by Mark-Jason Dominus'
  digging skills) POSIX compliance.

In Brief
  Rafael released a new snapshot of perl 5.6.2.

  Stas Bekman asked why the PERL5LIB environment variable is ignored when
  taint mode is on.

  Autrijus Tang added support for signing distributions in

  Richard Clamp is working on the perl bug triaging effort, via the
  bugmongers mailing list. He posted an announcement draft :

  Following a suggestion by Alex Hudson, support for Linux abstract Unix
  domain sockets were implemented. Those are basically like regular named
  Unix domain sockets, but their name begins with a "\0", and they don't
  live in the filesystem.

  David Dyck found that the perl parser was fooled by non-existent
  filetest operators (such as "-F" for example.) (bug #24212.) This was
  fixed by Rafael.

  And, Perl 5 turned nine years old.

About this summary
  This week's summary was once again written by Rafael Garcia-Suarez.
  Summaries are published weekly on and on a mailing
  list, which subscription address is
  Corrections and comments are welcome. Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About