The Perl 6 design team met by phone on 06 August 2008. Larry, Jerry, Nicholas, Jesse, and chromatic attended.
c:
- started the release management expectation for post 1.0
- want to get that nailed down before the end of the year
- applying patches, fixing bugs
- more small projects, the better
- helped Allison get the concurrency branch merged, looks fairly good
Jerry:
- agree with that timeline for getting documents sane on the release process
- lots of traveling for several Parrot and Rakudo hackers
- work has slowed down because of that
- the concurrency branch merge was pretty smooth
- able to dedicate more time to Parrot and Rakudo lately
- reviewing Kevin Tew's GSoC project as it nears completion
- running into a miniparrot build problem
- trying to help him debug that
- taken over mentoring for Adrian while Moritz is on vacation
- he's been pretty quiet this week
Larry:
- the standard grammar parses 99+% of t/
- four of the six failing tests are the fact that it does not install new grammar rules for user-defined operators
- one of the tests fails because the author of the test thought that Perl 6 as should read your mind as to where to apply adverbs
- straightening out some of the parsing snafus
- we need to be able to distinguish between types and variables all over the place
- now
:: really isn't a sigil - it gets in the way all over again
- special symbols like
::?CLASS are specially parsed at this point - responding to questions on p6l about exception handling with defined-or and MMD semantics
- mostly just hacking on the standard grammar and the tests
- always have to decide whether the test or the grammar is wrong, when they conflict
- thankfully it's often the tests
- going to Copenhagen next week
Nicholas:
- part of the problem of Perl 5 and the CPAN gets solved with Perl 6
- which allows different versions of modules by different authors
Jesse:
- had several chats with Audrey, who's hacking on Perl 6 again
- she's taking Pugs apart into separate useful packages of Haskell code for Hackage
- it's a Voltron-style development method for Pugs
- the current version passes about the same number of tests as the version of two years ago
- it now passes them in 15 minutes, instead of two hours
- a dramatic performance improvement already
Nicholas:
Jesse:
- some internals changes
- she's documented some of them
- she's waiting for the 6.10 version of GHC, which adds built-in DSL support
- it'll be a lot easier to port the standard grammar for running in Pugs atop GHC
- it's easier to drop pure Perl 6 implementations of Perl 6 on top of Pugs and get them to work then
- now a lot of cleanup and stabilization is going on
- she wants a lot of people hacking on Rakudo, v6.pm, and the tests
- also Haskell hackers
- Flavio's back to hacking, but we haven't caught up
- there's a lot more activity now than three weeks ago
- I'm presuming there's no call next week, with so many people at YAPC::EU
Nicholas:
- applied a patch to get parallel tests working in the Perl 5 core
- running full tests in parallel is only slightly faster than serial
Jesse:
- use
-j 9 on a dual-core machine with the new Test::Harness -
--fork also works well -
--save and --slow front-loads the slowest tests, which causes things to run faster
Nicholas:
- Andy Armstrong says this is moderately brittle
- I think he pays more attention to issues from p5p at the moment than Parrot
