All the Perl that's Practical to Extract and Report

Perl 6 Design Meeting Notes for 21 March 2007

chromatic on 2007-03-28T18:36:43

The Perl 6 design team met by phone on 21 March 2007. Larry, Damian, Allison, Nicholas, and Jesse attended. These are the notes.

Larry:

  • Paddled as hard as I can to stay in one spot.
  • Been thinking about the synopses updates that are not yet in the repository.
  • Still thinking about the autotopicalisation issue.
  • I've an invite to Iceland in the middle July, to a functional programming conference, to be the representitive of non-functional programmers.
  • Need to figure out if I can get to that - not cheap to fly to Iceland, but the conference is invitation only, roughly "top 25 functional programmers in the world"
  • Audrey says she'll go if I go. I'll go if she goes...

Allison:

  • Apply to TPF for a travel grant.

Larry:

  • Watching the MAD changes in perl5-porters.
  • Brushing up on my Haskell again - third or fourth time learning the language.

Jesse:

  • Using it to hack Pugs?

Larry:

  • No, not that good at it yet. I've made a few tweaks, but that's about it.

Jesse:

  • A friend commented that in Haskell, Audrey's code looks much like anyone else's code. Maybe Haskell enables all programmers to think like Audrey.

Larry:

  • Look at any particular function and it will be a brain pretzel.
  • Partly it's deceptive due to weird things about their syntax.
  • What are the semantics that they really want here? How can be provide those in a user accessible fashion that does not melt people's brains?
  • If we provide them.
  • Looking at features in Haskell and thinking "That's just one of these in Perl 6; that's just one of those in Perl 6"
  • There are very few things in Haskell that we don't have in Perl 6.

Jesse:

  • To see if Haskell can be translated to Perl 6?

Larry:

  • They're coming at it from very clean semantics.
  • eg latest round between Jonathan and Damian S9 on enumerated indexes,
  • Haskell has it as a type, rather than an enumerated array.
  • Should that notion be transferred over to the type of the array, rather than the declaration? But maybe it's something that Haskell does but Perl 6 doesn't need to.
  • Going to Rhode Island to talk at Brown University. Then taking a week off - not sure if there will network access on holiday.

Damian:

  • As Larry mentioned Jonathan and I have been back and forth on user defined indices for arrays. I think we're pretty happy with it now.
  • In response to a request from perl6-lang put out an XHTML version of the pod synopsis, S26. That required the last part of the implementation that previously I hadn't got a round tuit for. That perl 5 modules should be up on the CPAN reasonably soon. [Damian notes that real life may rudely interrupt him again]
  • Apart from that, read synopses, send feedback to Larry.

Allison:

  • Had to cancel this week and next week's trip, which happens to give me more time to hack.
  • Will made the Parrot 4.10.0 release, and named it "the release formally known as "prince""
  • It seemed to go smoothly, tracking it on IRC, but there are still a few kinks there.
  • Working on patches.
  • Big edit on the Parrot FAQ, took out Perl 6, should be moved to languages/perl6
  • Finishing up the OO PDD, which contains notes from Jonathan, and from talking with Particle. Just to check it for consistency, clean up the languages, and then can call it "implementable".
  • Hopefully by next week I can move it out of the drafts directory.

Jesse:

  • Patrick?

Allison:

  • He's been absorbed by family life this last week.
  • We're about to reach the point where we need to draw in more volunteers to work on Perl 6 on Parrot.

Jesse:

  • Ann would like to transition summaries to someone else, so I'm looking for someone to take it over. If you know someone willing, please tell me.
  • The Pugs IRC channel has some talk about distributed summaries
  • Leon and I have a draft announcement for Perl 6 microgrants.
  • It's with Ovid and Bill for blessing, then will post it.
  • Will be around next week, not 100% sure about the following week (Japan has the bandwidth and I have Skype, so if I'm awake it should be OK)

Nicholas:

  • Fixed the (genuinely) last test failure in MAD - the abort() code for local $^L, and fixed the PERL_DESTRUCT warnings about a leak - turned out to be in dead code elimination on conditionals on constants.
  • Noticed that Gerard has sent more MAD patches to perl5-porters. I commented on a short patch where I thought it should be conditional on PL_madskills at run time rather than PERL_MAD at compile time.

Larry:

  • 2 level of protection, PERL_MAD and madskills

Nicholas:

  • That was the one I spotted quickly
  • [Basically, when PERL_MAD is enabled at compile time, it compiles in run time checks on PL_madskills, which allow the XML dumping to switched on, but the default is still to parse, compile and run programs as normal, albeit more slowly due to the runtime checks]

Jesse:

  • Can you give an overview of whether the patches are right?

Nicholas:

  • Not really. Whilst I merged the MAD changes it, to me it was an exercise of spotting patterns in code. The MAD code effects the optree generation, which is not something I'm familiar with. I can't give any sort of quick response - I'm out of my depth here.