All the Perl that's Practical to Extract and Report

Perl 6 Design Minutes for 19 December 2007

brian_d_foy on 2007-12-21T01:41:00

The Perl 6 design team met by phone on 19 December 2007. Allison, Patrick, Jerry, Flavio, Richard, and chromatic attended.

Patrick:

  • the new perl6 compiler now passes all of the 01-sanity tests again
  • so it's not the new perl6 compiler
  • it is now just the perl6 compiler
  • the number of patch submitters is also increasing
  • Jonathan Worthington added in a preliminary implementation of junctions, with no prompting from anyone
  • he also added in hash subscripts and a few other things with minimal prompting
  • Flavio Glock submitted a patch for BEGIN blocks
  • I'm pleased with the way the new compiler is structured
  • it appears easy for other people to add things
  • next up is working on the test suite
  • also writing more documentation on how the compiler works and how the pieces fit together
  • helped Jerry get Punie running, at least behind the scenes
  • a few little things here and there
  • I look forward to getting more done this week

Jerry:

  • made Punie run on PCT for the 20th anniversary
  • Jeff Horwitz made Perl 1 and Perl 6 run in Apache via mod_parrot

Patrick:

  • 20 years of Perl has brought us mod_punie

Jerry:

  • I improved some Parrot error messages to make life easier
  • concentrating now on getting Test.pm running so that we can use the Pugs test suite

c:

  • fixed some bugs in the tests, to try to help the release go more smoothly
  • hope to remove a bunch of removable deprecated stuff
  • tried to remove src/objects.c, but there were a couple of used things in there
  • had a discussion with Adam Kennedy about making a perl6 binary, probably with the pbc2c utility and installing the PBC files appropriately
  • seems like a nice little afternoon's diversion
  • could even make a package for it for Debian or something

Richard:

  • I had cake
  • devoting my efforts into two places
  • the chaotic maelstrom which is the Perl 5.10 press release
  • we'll converge on a solution for that shortly
  • the other bit has been doing some donation prospecting, planning, and contact

Patrick:

  • is anyone writing a report from the recent Chicago Hackathon?

Richard:

  • I haven't had any contact with them

Allison:

  • finished up the Events implementation
  • plan to enable it today
  • finished a draft of the MMD PDD; I'll commit that shortly

Flavio:

  • started talking with Patrick about implementing some things that kp6 needs
  • we want to bootstrap kp6 on Parrot
  • we could run it on P6 on Parrot, then use it to compile itself
  • we ran into some problems, but Patrick is looking into them
  • also working on kp6 on Perl 5
  • bootstrapping will be difficult there, because the speed is so low
  • we have alternate options
  • perhaps we could make it work over Moose
  • then we have to give up on some features

Patrick:

  • I sent a message about the eval problem we may have
  • I'm starting to look at running the Pugs tests
  • they need a lot of reorganization
  • many of the tests use eval to do things that we don't need eval to do
  • I can't tell if they're necessary, or if someone threw it in there because they could
  • I don't want to interfere too much with what Pugs is relying on

Flavio:

  • the main reason for eval was to work around errors that prevented the tests from running
  • I think it's okay to remove
  • maybe add some eval extra tests just in case
  • if it's testing eval, not just the actual test

c:

  • I agree completely
  • I'd hate to break the tests for Pugs though, especially if there aren't people hacking on the Haskell core often

Flavio:

  • there aren't really

Patrick:

  • expecting eval to work early on seems like kind of a stretch for a new implementation

Allison:

  • using it to protect yourself from code that doesn't parse, for example

Patrick:

  • right
  • also looking for a way to mark tests as TODO for various tests
  • we looked at a way of using smart comments for that
  • Larry had an idea about that

c:

  • seems like a data-driven test
  • much like the Parrot tests driven by Perl 5
  • you could do that for just the minimum features necessary to catch parse fails

Patrick:

  • I'll try the comment approach for now and see how it works for Perl 6 on Parrot
  • might generalize later
  • is the current subdirectory organization of the tests in Pugs decent, or does it need reorganization?
  • I find myself looking through various places for the right tests
  • I don't know if that's me or not
  • maybe organizing them by Synopsis makes sense

Jerry:

  • I agree
  • test reorganization has been on the list of things we've wanted to do for a while
  • people are available
  • we haven't had a direction for that yet

Flavio:

  • many tests were added randomly
  • there's a site that links tests to the specification
  • but moving around the tests won't break the documentation, so it's okay

Patrick:

  • we could merely duplicate them for a while
  • keep track of the duplicates
  • then that site could switch over to the new ones

Flavio:

  • you can just move them around
  • you don't have to ask permission
  • might be good to start with a stripped-down Test.pm and then an advanced one

Patrick:

  • we have an intermediate one
  • maybe we can segregate tests based on which module we can use

Flavio:

  • any thoughts on logging the #parrot channel?

Allison:

  • we discuss that once in a while

Patrick:

  • I remember people had concerns about that

Allison:

  • mostly I just run screen
  • no call next week
  • let's plan to talk on the 2nd then