This Week on perl5-porters (31 March / 6 April 2003)

This Week on perl5-porters (31 March / 6 April 2003)
  Patches, crashes, hashes and stash caches, these are a few of my
  favorite things. If you like them, too, this week's P5P summary is for
  you !

Stash Cache
  Arthur Bergman has the idea to use a cache to store symbol tables
  (a.k.a. stashes) for class method calls, to avoid looking them up from
  the package name each time they're invoked. He proposes a patch, with
  which he observes an average 180% speed increase (on his system).

      http://xrl.us/fm9

Patches
  Ilya Zakharevich submitted a load of patches to maintperl, some of them
  related to the Perl port to OS/2, some of them more generic, to
  MakeMaker, Net::Ping, Time::HiRes. He also posted some comments about
  the ithread implementation in 5.8.0, and tried to have microperl
  working. (Look in the perldelta manpage for perl 5.8.0 if you haven't
  heard about microperl before.)

      http://xrl.us/fna : threads goofs
      http://xrl.us/fnb : microperl

  And that's not all. He also proposed to add built-in support for the
  now-standard "--help" and "--version" switches to Getopt::Std. This is
  followed by a discussion with Johan Vromans about extending this feature
  to Getopt::Long.

      http://xrl.us/fnd

In brief
  Alberto Accomazzi reports that perl 5.8.0, compiled with the
  "usemymalloc" configuration option, doesn't seem to be able to handle
  more than 1GB of RAM. This being most probably due to an inner
  limitation (although nobody commented on this).

      http://xrl.us/fne

  Bug #21765 is about split() not returning the desired kind of
  false/empty value, when called in a specific assignment context.
  Yitzchak Scott-Thoennes sent a fix. Demonstration here :

      http://xrl.us/fnf

  Bug #21787 reports a problem with the "fields" pragma. Michael Schwern
  says that this looks like a bug which was just reported in Class:Fields,
  which includes its own version of fields.pm.

  Michael Schwern, busy on the MakeMaker front, asked how portable the
  redirection construct "2&>1" was, when used from inside the qx//
  operator. Apparently it's portable on platforms that provide a shell
  that understands it, on OS/2 (that provides an emulation), and on VMS,
  but only when the external command run by qx// is perl.

  Steve Hay provided several fixes to the installhtml utility, that
  generates the perl standard documentation in HTML format, and that was
  severely messed up.

  Chip Salzenberg wonders, lonely, about the FIXME comments that can be
  found here and there in the sources.

  Sadahiro Tomoyuki released new versions of Unicode::Normalize and
  Unicode::Collate.

About this summary
  This summary was brought to you by Rafael Garcia-Suarez. Weekly
  summaries are available on <http://use.perl.org/> and/or via a mailing
  list, which subscription address is [email protected].
  Comments, corrections, additions, and suggestions are welcome. Apologies
  to Rodgers and Hammerstein.