Perl 5 Porters Weekly: June 24-30, 2013
Welcome to Perl 5 Porters Weekly, a summary of the email traffic of the perl5-porters email list.
Topics this week include:
- Idly musing about vtables for hashes
- Raison d'être of op_ppaddr
- Building real CPAN dists in core perl
- postfix dereference syntax
- experimental: linux abstract domain sockets
- the mystery of undef upstreams
- Perl 5.18 and Regexp::Grammars
- What does "deprecated" mean?
- RFC: Any objections to preserving spaces in prototypes
- Removing old utilities
- podcheck and long lines
Idly musing about vtables for hashes
Steffen Mueller muses about a pluggable implementation for hashing with the potential benefit of having multiple hash-like structure implementations that could be swapped on a per-instance basis - this could tighten up memory utilization of blessed hashes used as objects.
Raison d'être of op_ppaddr
Steffen Mueller riddles P5P about the ppaddr opcode. Paul Johnson explains that he uses some of it for evil purposes in Devel::Cover (and there are some other contributors to this answer.)
Building real CPAN dists in core perl
LeoNerd asks about making dual-lifed modules build more like a traditional "from-CPAN". David Golden replies that he's in favor of this generally speaking.
postfix dereference syntax
RJBS posts some of the ugly inconsistency in perl's dereference syntax. He wonders how it could be improved but no one seemed to respond.
experimental: linux abstract domain sockets
Rik posted another "experimental" part of perl from 5.12 perldelta. He's leaning toward removing its experimental designation. There was no discussion.
the mystery of undef upstreams
The fearless pumpking Ricardo Signes posted a message that he'd like to clean up the Porting/Maintainers.pl file in two ways:
- determine the upstream for undecided dists
- drop the "person on the spot" listing for things where it's not useful
Discussion ensued.
Perl 5.18 and Regexp::Grammars
Perl 5.18 broke the Regexp::Grammars module and there doesn't seem to be an easy way to make it work. This thread is discussing what can be done to make it work again in the future. Also it appears that there was a bug that surfaced in the 5.18 regex engine which Dave Mitchell isolated and Karl Williamson was looking for a suggestion on what ought to be done to fix it.
What does "deprecated" mean?
There's been a thread running since June 17 about whether the use of the
bare literal << to mean <<"" should be marked as deprecated (with
a warning) or should be a hard error or nothing should be done (business
as usual.)
This week the thread turned philosophical with some of the people in the thread trying to hash out what the word "deprecated" means for an end-user and what it means in terms of future work for P5P.
RFC: Any objections to preserving spaces in prototypes
Peter Martini asks if anyone objects to preserving whitespace in prototypes. The current implementation strips whitespace.
Removing old utilities
Leon Timmermans proposes to remove several old utilities from the core distribution. In particular he suggestions removing these:
- s2p
- a2p
- find2perl
- h2xs
- c2ph
- h2ph
- enc2xs
- libnetcfg
- pl2pm
Is anyone still using these, and if so, should they be in the core distribution?
podcheck and long lines
RJBS reported the status of his branch with some work on the podcheck/long lines controversy of a few weeks ago. He writes that he stalled on this but Brian Gottreu picked up the work and submitted a patch which was applied.
Weekly p5p summaries