This Week on perl5-porters (3-9 May 2004)

On the menu of the P5P summary this week, you will find language constructions, segmentation faults, proposals for new tied methods, pronunciation issues, and (in fine) a few bugs.

More on last

Continuing a thread from last week about the last statement used to exit an eval() block: last and next, when used to exit from a subroutine, not only exit, but also branch to the specified loop/label of the caller. This is definitively a feature, since (as Tim Bunce remarked) Test::More::skip() uses this, and (as Hugo van der Sanden pointed out), Apocalypse 12 uses this to define method lookup strategies.

    http://groups.google.com/groups?selm=200405051134.i45BYET26083%40zen.crypt.org

open pragma segfault

Tomasz Pala reports (bug #29437) a case of recursion in perl's character encoding routines, which obviously (once diagnosed) leads to stack recursion and to segmentation fault. The sample code involves somehow the open pragma.

    http://groups.google.com/groups?selm=rt-3.0.9-29437-87042.12.9753390677264%40perl.org

Truncation of tied filehandles

Dan Boorstein proposed a patch against bleadperl to add a TRUNCATE method to tied filehandles. Tassilo von Parsval commented that this new method should probably be made optional, for simplicity and compatibility.

    http://groups.google.com/groups?selm=20040508103121.519a8d9f.dan_boo%40bellsouth.net

More pack() work

Marcus Holland-Moritz, who implemented a couple of weeks ago new modifiers (< and >) for pack() templates, added a way to group these with parentheses.

    http://groups.google.com/groups?selm=20040503201441.1b058e0d%40r2d2

Weak references problems in 5.8.4

Stas Bekman reported that Scalar::Util::weaken() (used to make weak references) has new problems in perl 5.8.4 (as it attempts to free unreferenced scalars, unveiling a deeper malfunction.) As those problems don't appear in bleadperl, Dave Mitchell does some investigation about the patches which caused it, and proposes a solution -- a patch from blead to integrate in maint.

    http://groups.google.com/groups?selm=4097EF6F.8050401%40stason.org

How do you say !~?

Yitzchak noticed that using !~ on an uninitialized value reports the warning as having occured in a not operation, which is not exactly the case, and proposed to change the message; which raised the question, what is actually the name of the !~ operator?

    http://groups.google.com/groups?selm=20040509075919.GA3752%40efn.org

In Brief

Dan Dascalescu reported (bug #29346) that passing an undefined argument to the int() or abs() built-ins produces the Use of uninitialized value twice. Another history of magic handled twice.

H. Merijn Brand is back on Cygwin smokes. This time, there were some IPC::SysV failures; Yitzchak Scott-Thoennes pointed out that a service needs to be started on Cygwin to make them work.

    http://groups.google.com/groups?selm=20040506142843.98DB.H.M.BRAND%40hccnet.nl

David Muir Sharnoff reported that Scalar::Util::refaddr() (a function that returns the memory address of the reference it gets passed) doesn't work when magic is somehow involved (bug #29395). Marcus fixed this.

About this summary

This summary was written by Rafael Garcia-Suarez. Weekly summaries are published on http://use.perl.org/ and posted on a mailing list, which subscription address is perl5-summary-subscribe@perl.org. Comments and corrections welcome.