Larry:
- I took Geneva to school last week
- since then, I've been playing with a new IBM ThinkPad I received as a
gift
Patrick:
- in addition to dual-booting, have you considered running VMWare?
- maybe I can get you a copy
Larry:
- I haven't used it since it became more corporate
Patrick:
- I use it quite a bit on my systems
- being able to copy and restore VMs is quite nice
Larry:
- thinking about everything people talked about on p6l in my absence
- have expressed opinions about some of them
Luke:
- your points about Yuval's type inference is quite valid
- I can't understand the stupid error messages
Larry:
- "Something's wrong at line 42"
- about that useful
- also thinking about tree transformations and two-way transformations
- seems to be one of Autrijus' pet ideas: two-way transformation
- currently, if you use tree transformations to do macros, he wants to
turn that back into unmacroized whatever
- you could write hygienic source filters that way, as it were
- notion of unifying warnings with exceptions
- ... where a warning is a guaranteed restartable exception printable by
the outermost layer by default
- seems workable
Patrick:
- I like that
Damian:
- I like that
Luke:
- you've just pulled continuations out into user space
- you promised you wouldn't do that
Larry:
- not overtly
- do it by running CATCH closures up the stack and keeping the stack and
committing to running back when you catch something
Luke:
- that's a continuation
Larry:
- we could hide that with .resume or something
Damian:
- it ought to be a separate kind of named block
- maybe should be able to catch it in CATCH or something
- probably cleaner with a WARN block
Larry:
- possible that they're a type of control
Damian:
- a lot of people will want to catch warnings
- it'd be easier to capture it in a name that self-documents
Larry:
- the name of the warning may be self-documenting anyway
Luke:
- Piers has always wanted interactive editing
- if an exception comes up to the top level, it can say "Something's
wrong; fix it and I'll keep running!"
Larry:
- that's part of what I was thinking
Jesse:
- did you have a good vacation, Damian?
Damian:
- I wish it were a vacation
- tore maybe half of the muscles in my back
- had a fair bit of physiotherapy over the past couple of weeks
- not good for sitting *or* lying down and doing stuff
- also had a hard disk crash on my primary laptop
- fortunately had backups, but they don't always back everything up
properly
- been interesting times, especially being two weeks away from leaving for
EuroOSCON
- I have been creating annoyances on p6l, as you've seen
- very nice to have Larry back stopping some of the madness
- very little progress to report elsewhere
Jesse:
- Patrick brought up something in the implementor's meeting
- S05 is out of date
Damian:
- it's waiting on me to put in stuff about capture variables
Jesse:
- are you the right person to be doing it, or should it go to someone
else?
Damian:
- yes to both
- I have a completed version nearly done
- it's just slipped
- I'll get onto it in the next day or two
Patrick:
- have done more reading, but not much implementation
- housing people for the past week or so
- Leo and Chip and the Parrot folks have merged the branches
- makes my life much easier
- very happy about that!
- can move forward on PGE again
Luke:
- had a hard drive failure, like Damian
- if you have a RAID, replace a failed hard drive
- I didn't replace one hard drive after it failed a month and a half ago
- then another failed
- it didn't do much for me
Patrick:
- it gave you an extra month of uptime
Jesse:
- also Autrijus recommends that you make sure that your RAID controller
does not format *all* of your disks when you add one
- anything new?
- mostly hacking tree transforms?
Luke:
- wrote a new module, Language::AttributeGrammar
- implements attribute grammars
- it was easy to write
- does everything UUAG does (reference implementation)
- trying to make it work nicely with PPI and Parse::RecDescent
- presumably usable to implement something real
- pretty much finished theory theory
- it's in the Pugs repository
- some lambdacamels have reviewed it positively
c:
- how difficult is it to port L::AG to PIR?
Patrick:
- I was thinking that too
Luke:
- hm, somewhat difficult without a Perl 5 preprocessor
- lots of lazy evaluation and nested closures
- also thinking about use fatal a bit
- my first fatal would be semi-fatal
- things return undef if they fail, but if I ignore that undef things fail
Damian:
- it returns undef but... ?
Luke:
- if I ignore that, it causes my stuff to fail
- behaving like an exception
Larry:
- fatal in void context, basically
Luke:
- a fatal undef in void context makes it fail
Damian:
- that turns out not to be useful; PBP talks about that
- fatal in p5 has a :void mode
- it's not that useful
- you want it to be fatal in non-boolean context
Luke:
- you'd get a fail back in void context
- you assign to a variable and that's in void context
Damian:
- that's not how context works
Luke:
- there's void context for the assignment expression
- that's the semicolon
- I should write this up
Damian:
- return failures in a non-boolean context should raise the exception
Luke:
- does // give that context?
Damian:
- maybe it's not the only context
- things like assignment must be not considered void
- write it up and we'll talk about it
Luke:
- also thinking about junctions
- not coming up with any happy medium
- I'll keep it on the list
- have a meeting with a professor here
- he's done significant research in attribute grammars
Allison:
- grab papers if you can
- they're very valuable
- they're rarer than hen's teeth
- my biggest idea has been writing a weblog about "This is why we really
need Parrot"
- I've been doing a lot of reading about CLR and Mono and things like that
- at Amazon.com, I think I managed to kidnap a former CLR developer whose
NDA will expire shortly
- Autrijus desperately wants me to check in the compiler tools notes as
they are
- I want to make a few changes, but I want to do that in the next few days
Jesse:
- Chip was at the implementors meeting
- he's settled, working, and his internal clock has reset
- we talked about PDDs and how to move faster there
- he'll start by reviewing their organization
- then post the index of what he's writing in his head
- then he'll sketch out each document and get some movement on them there
- lots of discussion of necessary work to merge Leo's branch
- the merge has happened
- I've been getting questions about contributor agreements
Allison:
- there's a new update
- the final thing is review of the patent clause
- especially in relation to the clause we put in the license
- we can't take either in isolation
Jesse:
- what do we need to get the review?
Allison:
- Roberta says we'll have a revision in mid-to-late October
Jesse:
- okay
- that's it for me at the moment
- who'll be on the Perl Whirl?
Allison:
- Larry and I
Patrick:
- I won't be here either
Jesse:
- let's postpone it until next week
- there will be no call on the 5th, but one on the 12th
