The Perl 6 design team met on 19 September 2007 by phone. Larry, Allison, Jerry, Nicholas, Jesse, and chromatic attended.
Allison:
- launched the PMC PDD this weekend
- first milestone in the new timeline
- hacking madly on the object branch since then
- it's so much easier on a branch; I can't tell you how much
- fixed all but one problem with PGE and TGE
- PGE has a conflict in the
new method that Patrick will work on
- fixing other code throughout the repo that relies on assumptions of the old object system
- or skating around assumptions in the old object system
c:
Allison:
- it's going pretty fast
- I hope to merge soon after Patrick fixes his changes
- yesterday I gave a Parrot talk to the Seattle Perl Users Group
- they were an awesome audience
- good talk
Larry:
- thinking about the semantics of self-optimizing grammars
- self-lexing grammars
- otherwise incorporating a lot of suggestions into the specs from various people
- particularly Wolfgang Laun
- there's no such thing as a long dot anymore
- he has more suggestions on S06
- the parameter syntax that gives a thing two different names probably shouldn't
- a variable's official name in that case should be the adverb
- you can stack those to give it multiple names
- then you could essentially have an anonymous parameter name
- I think that's a good idea
- posted a long article on PerlMonks about the extensibility of the standard grammar
- response to a post by blazar
c:
- we released Parrot 0.4.16, mostly thanks to particle
- I'm still tracking the Tcl x86/Darwin segfault
- unfortunately, it makes gdb crash
- I need to review the PMC PDD
- talking to Kevin Farnham about Intel's Threading Building Blocks
- he's looking at that and I'm talking to him about helping
- could give us a type of concurrency or an option for concurrency in Parrot
Jesse:
- Adriano Ferreira is writing a series of mini-articles on Perl 6 operators
- he's doing well at roping in the Perl 6 community to help revise and edit
- they'll end up under an open license so we can include them in project docs
Nicholas:
- someone wrote to p5p:
open( state $sh, 'filename' )
- the open happens lots of times
- would the magical
state make a START block work on that?
- what happens to
state variables in lvalue contexts?
- is there anything analogous in Perl 6?
- Rafael thought it might only affect
tie, open, and sysopen
Larry:
- it's just a weird form of an lvalue context
- it ought to work in an lvalue context
- de-sugared Perl 5 might have trouble with that
- but from the standpoint of the surrounding syntax, it really ought to be trasparent when it gets a reference back what its initialization semantics are
- in
open state $fh, the initialization is not to initialize it subsequent times
Nicholas:
- in that case, it's correct for the
open to be inside the START block
Larry:
Nicholas:
- someone wrote and said that it should be
Larry:
-
state does not propagate its semantics outward
- just to an explicit equals
Nicholas:
- there's still an open question about list context
- we detect that and suggest we may fix it in a future version
