Where does the time go? I blame folk festivals. Once I'm getting busy with the teacher training I'm going to have weekends free to write the summaries. And if you'll believe that...
We start with perl6-internals.
Discussion of implementing incremental garbage collection continued.
Mattia Barbon asked for some clarification of how the calling conventions work for unprototyped subroutines in IMCC. Specifically, whether IMCC was behaving as documented in the Parrot Design Document (PDD3). Leo reassured him that the TODO entry that had caused his disquiet was obsolete and that IMCC does the right thing.
Steve Fink asked how to go about returning something from dynamically compiled code. He, Leo, and Dan went over the issues. An unnamed (and as yet unvolunteered) volunteer will probably write it all up and add it to the parrot compiler FAQ.
Stephane Peiry and Leo continued to work on getting NCI callbacks working with the GTK library.
They finally tracked the problem down,
but it looks awfully like Stephane's going to have to re-implement the
gtk_main loop to get everything working.
Best of luck Stephane.
ParrotGTK would definitely be a cool thing.
Robert Spier announced the creation of Planet Parrot, an aggregator of Parrot related blogs. If you have such a blog, Robert would probably be interested in hearing from you.
Leo and Dan continued to discuss and work on documenting Parrot's memory management API.
Dan noticed a problem with garbage collection and NCI PMCs. Leo tracked the problem down and fixed it.
I love it when a plan comes together.
Base behavior is disgusting and should be stamped out I tell you. We have a moral duty to...
Actually, the subject refers to Dan and Leo's desire to nail down the behavior of Parrot's basic types. He outlined the issues and asked for discussion. And that's what he got.
It looks like Parrot's going to have a reasonably complete set of numeric types which will allow accuracy nerds to avoid reals for as long as possible. Which will be nice.
Comfortably winning the "weirdest subject line of the week" award, Sam Phillips laid out some of the problems he's having implementing a Basil compiler that targets Parrot. Leo supplied answers.
As well as discussing basic types, Dan opened a discussion of the proper semantics of Parrot's low-level maths (yeah, he spelt it "math," but I'm English and I'll spell it correctly) operators. In particular he proposed variants that throw exceptions if strange things happen.
Joshua Gatcomb posted some patches he'd made to F<parrotbench.pl> to make it a little less fragile and more useful. A few days later he announced a slightly more finished set of patches (Leo had checked his original patches in rather sooner than expected).
Dan Sugalski looked upon mod_parrot and saw that it was cool, to the point of being laid out on a mortuary slab. He thought that the time had come to try and put some life back into the project, now that Parrot's so much more capable. He asked for volunteers and Jeff Horwitz seems to be (the only?) one of 'em.
Leo posted another of his discussion documents, this time on instantiating PMCs. No response so far.
In the usual freewheeling perl6-language way, a thread entitled "Return with no Expression" ended up with a discussion of how to disambiguate a
::, which is part of
... ?? ... :: ... expression and
:: in its role as module sigil. (According to Luke it's easy; white space is required.)
Discussion of the draft synopses continued.
Aaron Sherman wondered if Perl 6's incarnation of POD would have support for tables. Luke wasn't overly keen on Aaron's proposed syntax, and Larry pointed out that they'd managed to use Perl 5's POD and some clever processing to write tables for the third edition of Programming Perl. Once again, it seems that everyone agrees that POD is great, but could be greater, but nobody can agree on how to make it greater. Me, I think Larry got it right enough.
Ovid had some questions about how Constructors work in Perl 6 and about how to disambiguate methods when using Roles. Larry and Damian provided answers.
Instantiation was on Aaron Sherman's mind, too. He proposed some syntactic sugar for creating objects at "use" time. For some reason, syntactic sugar proposals do seem to create lots of discussion.
I am reliably informed that a lot of last week's links were broken. Which is odd because the code to generate them hasn't changed. It looks like the problem's down to the gateway between nntp.perl.org and groups.google.com being slightly radished at present.
However, I'm going to stick with the Google links in this summary in the hopes that normal service will be returned shortly and things will get back-filled. If anyone has a handy tool for getting from a message id to an archive URL that works then I'd be happy to hear from them.
If you find these summaries useful or enjoyable, please consider contributing to the Perl Foundation to help support the development of Perl. You might also like to send feedback or contributions to a "Getting Piers to OSCON 2005" fund to mailto:email@example.com.
http://donate.perl-foundation.org/ -- The Perl Foundation
http://dev.perl.org/perl6/ -- Perl 6 Development site
Amazingly, there's a small amount of newer content on my web site this week. Will wonders never cease?