This Week on perl5-porters (22-28 December 2003)

This Week on perl5-porters (22-28 December 2003)

Due to holidays, this was a low-traffic week. That's one more reason to get it delivered in time.


Tels pre-released Math::BigInt 1.68, with new stuff, refactorization, memory savings, and new method aliases to make the interface more consistent. He also posted some benchmarks.


Enache Adrian re-optimized the implementation of the string concatenation with lexical variables; the previous optimization has been disabled because it was buggy.

Windows breakages

According to bug report #24731, the implementation of fork() has problems under Windows 2003 64-Bit on Itanium 2. Unfortunately, we lack porters with access to this kind of machine.

Speaking about Windows, we seem to have lost threaded builds of perl 5.8.x on Windows, for an unknown reasons. Investigations to follow... (Obviously, perl 5.8.3 can't be shipped without this being fixed.)

Other bugs

Ton Hospel reports that undef used with the range operator in list context is treated as numerical 0, even for a magical string range. Rafael fixed this.

Jarkko Hietaniemi reports that this small code snippet :

    *foo{CODE} ? 1 : 0

is a syntax error in perl. It shouldn't be. Rafael fixed this.

Marcus Holland-Moritz thinks he hits a bug in a recent development snapshot of gcc 3.4, because a sprintf() test failing in perl's test suite. The bug in fact turns out to be an old bug in gcc's optimizer, and is trigerred with the -O3 level since gcc 3.4.$ff6661e0$5700a8c0%40R2D2

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.