The Perl 6 design team met by phone on 20 August 2008. Larry, Allison, Patrick, Jerry, and Jesse attended.
Larry:
- decommuting from Copenhagen
- couple of tiring days wandering city
- mostly hacking on STD.pm grammar to speed it up a bit
- mostly propagating ratchet optimization around
- compiles itself in under 90 secs
- it's an improvement
- also working on adding rudimentary symbol table checking
- can figure out where it's misparsing things as function calls
- getting ready for trip to Japan next week
- lightweight languages conference, for some definition of "lightweight"
Patrick:
- returning home from Copenhagen, which was quite productive
- on the plane ride I implemented pieces for precompiled modules in Rakudo
- finished that up last night, so now Rakudo can use precompiled modules
spectest_regression on my box went from 12 minutes to under 4 minutes- it also means that use statement can start doing
load_bytecode things
- today been working on lazy lists with jonathan
- much of my time in the next couple of days will be writing reports and blog posts, catching up on events at OSCON and YAPC::EU
- also, at YAPC::EU hackathon Jesse and I put in some refactoring on Rakudo's ROADMAP, identifying task dependencies and some relative effort required
Jerry:
- GSOC update; mentors reviewing and submitting final reports
- looks very good that all three are completing
- very successful SOC
- we have good NCI stubs for Parrot, no longer need to write function signatures for things in C header files
- have 700 new tests added by Adrian for Perl 6 spectests
- have mostly complete GC implementation (some debugging needed)
Jesse:
- the recent blog post sounded a little "unhappy" with outcome
Jerry:
- yes, that the student's self-perspective
- however, it's design-wise complete, just not debugged fully
Allison:
- we didn't expect total rewrite of GC system
- we can do the GC integration with Parrot
- he did what we needed him to do, indeed he went much farther than I expected
Patrick:
- chromatic also mentioned that things are farther along than intimated
- just need a bit more debugging before we can pull it back into trunk
- didn't want to potentially block others' progress
Jerry:
- trying to catch up on work done at YAPC::EU
- much didn't happen online
- missed out on some good conversations
- today discussed exportation in Parrot on IRC with Patrick
Jesse:
- getting way too little sleep at YAPC
- very nice conversations with Patrick
- developed ROADMAP, put estimates of relative effort on them
- there's a lot less left to do than I would've estimated
- I also talked to Jonathan about what he'll be up to after his current grants expire
Allison:
- reverse commute from Copenhagen
- diving right into mmd branch, have tasklist
- Andrew and smash very helpful
- finished parsing of multi declarations
- build multi-dispatch table from C declarations
- then deprecate a bunch of C code to get to a single mmd impl
- roll that branch back into trunk
- should finish MMD milestone this week or next
- then it's onto IO implementation
Jesse:
- Perl 6 IO specification was flagged in Saturday's discussions
- Where does that belong?
Larry:
- I've looked at it a few times, it's certainly not complete
- "It's just a library issue!"
- it needs some TLC
Jesse:
- Are there others worth harassing for proto-TLC?
Larry:
- where IO interacts with concurrency I tend to sweep under the carpet
- I tend to not stress about it
- let's do one step better than P5 and not worry about it
Patrick:
- I also think that as long as we can soon provide an implementation that can do basic file I/O and sockets, then people can explore and play in the design space
Larry:
- as long as it's properly OO to begin with, there's lots of room to play
Allison:
- Parrot's IO model is becoming much more OO, yes
Jesse:
- could Rakudo take care of providing an OO interface to Parrot's IO?
Allison:
- independent of Rakudo, Parrot IO needs to be OO
- the old design seems to be "reinventing multiple systems to do the same thing"
- also wanted to mention volunteers working on test.parrot.org
- moving pages over to new parrot.org domain
- as soon as pages are moved over, we just need security certificate
