Here are the notes from today's conference call.
Larry:
- mostly thinking about the things Luke has sent out
- stuff on the mailing lists
- not much to report beyond what you've seen
- read the Scala paper
- my eyes glazed over in certain spots
Luke:
- it's a little dense
Larry:
- we have to find a different name than "data"
- no, you can't have vertical bar
- I think "data" is actively misleading to blue-collar people
- it's a little more abstract than that
- more like "a pattern of data"
Luke:
- a structure... a struct!
Larry:
- if you have a theory, then they're tenets, aren't they?
Luke:
- mostly thinking about the things I've been sending out
- started to implement some of the stuff I've proposed to see how it works
- tried to start as Perl 5 modules
- inevitably led me back to my type inferencer work
- I want strong typing without verbosity
- working on that again
- trying to get an algorithm for type inferencing that doesn't require
monotypes
- (I think they're too restrictive)
- it can check for errors
- not sure how to make it faster with the information it has, but...
- also think I used "theory" in two different ways in my post
- parameterized theory is more like a role or a multi-parameter interface
- the one from Scala is really more like a module
- basically trying to figure out how you extend a module and what makes
sense there
Larry:
- and if you poke multis into a role can you put them into a module?
- what we call them doesn't matter as much as what happens when you do
things with them
Luke:
- kind of want a global type for things
- for multis
- along with your operator-central language idea
- if you have a global type for plus that says "These things have to
implement the Num theory", then you can't override plus non-numerically
without fixing the global type
Larry:
- fits in with what Yuval was saying
- have your cake with performance and extensibility at the same time in
different lexical scopes
- I'm in favor of both performance and extensibility
Luke:
- who isn't, except for Java
Jesse:
- Luke, you were going to do summaries for these calls
- have you been doing that?
Luke:
- I did it twice and sent it to not-Piers but he didn't include it, so I
became lazy
Jesse:
- if we want to make them publicly available, you might be able to post
them to your use Perl; journal
- it takes 30 seconds to make an account
- we don't have to rely on the summarizers to include it
- at least we're talking
- seem to have managed to get the implementors in one virtual room weekly
for the past four weeks or so
- seemed to work best this past week
- but there was a fairly large queue of questions built up
- Chip, Patrick, Autrijus, Nick, Matt Diephouse, chromatic, Allison,
myself
- went through what people have done in the past week
- more importantly, talking about what they're sticking on
- mostly a result of having enough people say "I'm waiting for this from
that person" that I had the audacity to try to get these people in one
place
- much of this week was Nick tying Chip down and asking question after
question for Ponie
- Patrick mentioned that S05 is out of date with the state of the universe
Larry:
- I'm through S04 in the state of my revisions
Jesse:
- someone suggested goading Damian into updating this
- maybe it's you, Larry
- maybe it's someone else
Allison:
- at one point Damian said he'd do it
- maybe check and see if he's still interested
Luke:
- what about Patrick?
Allison:
- he doesn't have time
Larry:
- he's doing real work
Jesse:
- he says that if it continues to bite him, he might just cope
- he prefers that Damian, being responsible for the change proposal, do it
- I'll harass Damian today (via e-mail)
- I've sent out full transcripts of the sessions to sixperl
- people wanted unedited logs instead of trimmed minutes
- also talking to Nick about how to get Ponie moving faster
- trying to get some parallelizable Ponie tasks
- he feels that everything is on a critical path on what he needs to do
- one thing other people can do is to get Parrot signals in the right
place
- it's hard to embed Parrot and expect signals to do the right thing
- this needs a PDD
- don't know if anything has happened in the past couple of weeks with
Chip and contributing draft PDDs
- seems something Ponie-related that someone who is not Nick could do
Allison:
- is it possible to define someone to put up the templates?
Jesse:
- I mailed Chip over the weekend
- he's moving now
- I'll give him a bit longer before prodding again
- I don't know whether the wrong thing is wiki with authentication or
Subversion access or what
Allison:
- started hacking on an AST this weekend, written as Parrot objects
Luke:
- then you have a good idea of what you want to do with trees
Allison:
- yes
- I hadn't yet read that message
Jesse:
- did API changes come up at the London meet?
- we've run into this with RT every month and a half
- one of our dependencies changes its API
- people install old versions of RT with old "stable" versions of
dependencies
Larry:
- the long-term solution is that the version number is part of the name
Jesse:
- is there a medium-term solution that anyone has?
Jesse:
- seriously considering branching CPAN for distribution as part of our
tree
- I did not say fork
- we played with importing all of Backpan into an SVN repository
- it suffers from the same problem that search does
- it doesn't respect permissions on who can upload a module
- ends up slightly smaller than the tarred distribution of all of the
modules
