All the Perl that's Practical to Extract and Report

Perl 6 Design Minutes for 25 June 2008

chromatic on 2008-08-15T01:02:33

The Perl 6 design team met by phone on 25 June 2008. Larry, Allison, Patrick, Will, and chromatic attended.

Larry:

  • did YAPC
  • some hackathoning on both ends
  • gave a talk at Google Chicago
  • specwise, mostly clarifications
  • there's a new assertion in regexes in angle-brackets, <...> and friends
  • same function as out of regexes: not defined yet
  • a bare ... would match three characters in a row
  • renamed samebase switch to sameaccent
  • on the way to YAPC, I hacked in lazy lists, then spent the rest of the time debugging them
  • cleaning up the lazy switch languages and the memoization features which look for empty space and ends and such
  • no longer depend on single variables
  • an array checked by position within the current program
  • that array can have multiple memos attached to it
  • seems like it will work much better
  • the ws_ attributes are gone now
  • spits out a much cleaner tree
  • other than that, just chasing down lots of little bugs
  • added a Makefile
  • added the ability to write a different grammar and use a different mechanism from STD5
  • used to have to put derived grammars into STD5
  • don't have to do that anymore

Allison:

  • YAPC, hackathons
  • meeting with people and getting things done
  • back to debugging the PDD 25 branch
  • haven't been doing as much on as I hoped
  • spent the weekend with family though
  • looks like chromatic fixed a couple of things there yesterday
  • hoping to merge back in this week or next
  • that'll give us enough time to make sure it's safe for the next release

Will:

  • YAPC
  • Jim Keenan ran a Perl 6 and Parrot workshop
  • a lot of new people showed up and tried to build Parrot and the Rakudo binary
  • had a lot of feedback
  • had some patches
  • hopefully get some repeat contributors from that
  • we signed all of the paperwork for the Parrot Foundation last week
  • so it's all official now
  • no real coding
  • taking a look at the branch, trying to make sure Tcl works there
  • there are some PGE issues

Patrick:

  • Rakudo's regression spectests have 73 files, 943 tests passing (increase of 14 files, 238 tests from two weeks ago)
  • that's likely understating it
  • we haven't removed skip markers from lots of little fixed bugs yet
  • I have a tool which tells me what they are
  • expect that we'll pass over 1000 tests by this time next week
  • about 100 new passing tests every week
  • that's a good rate for now
  • YAPC was very productive
  • the hackathon and workshop were very useful
  • I have an impression of a lot more people starting to play with Perl 6
  • that may not be to Pugsian levels yet, but a lot more people seem to be coming in and saying "This doesn't quite work"
  • spent a lot of time this weekend looking at lexical handling
  • looked at the source
  • had to look at a lot of IMCC to figure out what's going on
  • doing some big refactoring of base classes and things in Rakudo
  • adding sanity, will help getting new features to work
  • continuing to refactor Rakudo's grammar to look more like the standard grammar
  • copying rule names and targets
  • eager to look at Larry's memoization changes
  • I had similar thoughts
  • that'll be good for all of the languages on Parrot
  • may be a standard part of the PCT grammar rules
  • removed some obsolete rules
  • the grammar can now process named zero-arity tokens
  • protoregexes and LTM will fix that better
  • more improvements to the test suite and the test harness for reporting progress
  • patched HLLCompiler to transcode from UTF-8 into fixed 8 when possible
  • it'll continue to use UTF-8 if it has to, but it'll take longer to parse
  • tired of people saying that Rakudo had trouble with Unicode, even though the problem was in Parrot
  • have a draft grant application for the Perl 6 grant, per the grant BOF at YAPC
  • Richard can start to move on it
  • will give other people an idea of how an application might look
  • starting to look at more PGE changes and improvements
  • want to add protoregexes and LTM
  • also looking at changes in pdd25cx branch
  • in particular, the new opcodes (peanut butter and bad beer)
  • local_branch and local_branch_return are okay
  • I might prefer local_return

Allison:

  • I figured something longer than three characters might work

Patrick:

  • there was some talk about people sticking invalid integers in the array
  • they think it's exploitable

Allison:

  • other places which do this check that the target is within the current segment

c:

  • that's easy enough to do

Patrick:

  • when I redo PGE, I might not even need these opcodes
  • they'll be useful now
  • but I don't want to invest too much time in them right now
  • the lexicals bug is slowing down Rakudo's progress
  • I have plenty to keep me busy
  • but fixing some of the underlying features in Rakudo depends on getting lexicals fixed
  • I can get some features to work by doing things one way
  • other features to work by doing it another way
  • but not simultaneously

c:

  • I can take a look at that

Patrick:

  • Jonathan said he'd look at that too
  • you two can coordinate on that

c:

  • applying patches
  • cleaning up IMCC
  • not sure it's worth much time revising it -- needs a lot of work
  • I'll spend more time applying patches, cleaning up code
  • going to branch for the Strings PDD soon
  • NotFound sent in a couple of patches
  • also mentoring Andrew and his GC work
  • he's mostly debugging now
  • I'm going to go through and clean up his code a bit
  • likely we can merge it back in soon

Patrick:

  • there's still a nasty design issue in lexicals and closures
  • it'll take me some time to describe it well
  • it's when we have a non-closure sub embedded inside a closure
  • if we do newclosure on something with an inner scope which does not do C, the pointers are all messed up