Stories
Slash Boxes
Comments
NOTE: use Perl; is on undef hiatus. You can read content, but you can't post it. More info will be forthcoming forthcomingly.

All the Perl that's Practical to Extract and Report

use Perl Log In

Log In

[ Create a new account ]

chromatic (983)

chromatic
  (email not shown publicly)
http://wgz.org/chromatic/

Blog Information [technorati.com] Profile for chr0matic [technorati.com]

Journal of chromatic (983)

Friday July 31, 2009
04:01 PM

Perl 6 Design Minutes for 10 June 2009

[ #39386 ]

The Perl 6 design team met by phone on 10 June 2009. Larry, Patrick, Jerry, Will, and chromatic attended.

Larry:

  • having some family business at the moment
  • reduced the precedence of adverbs from tighter than comma to tighter than item assignment
  • reduces the frequency of some FAQs
  • worked with Jonathan and others to nail down the notion that a named multimethod looks like a single object from outside
  • obviously multi from inside
  • if you declare a multi within a class and ask for a list of methods from outside, you only get that name once
  • discussed whether multimethods are interleaved like multisubs
  • they are not
  • they are only an implementation detail inside a given class
  • from outside the class, a method always appears to be a single routine
  • helps preserve encapsulation and sanity across language barriers
  • went over Pugs examples
  • ran them past STD; it coughed up a lot
  • fixed a lot of the problems
  • found a few infelicities in STD
  • a double-quoted string that ends with an interpolation of $foo., like at the end of a sentence, the parser sees $." and tries to find a method name in double quotes
  • the next double quote might be many lines away
  • the dotty op at the end of a double-quoted string now disambiguates against a method name in quotes if there's whitespace before the next double quote
  • it terminates the string
  • that can really send the parser off on a wild goose chase
  • converged the output of gimme5 and viv more closely
  • viv can translate more of the operator precedence now
  • gimme5 was overcompensating to add a semicolon at the end of a given block
  • that was a fossil from when it used do
  • now it uses 5.10's given/when
  • cursor's implementation of .cap to return matches in order was buggy and lost some matches from if/else statements
  • viv now translates :delete and :exists
  • that was part of the reason for changing the precedence of adverbs
  • adverbs are now parsed into a real AST node
  • treated more like postfix
  • modifying the node in place makes it difficult to maintain the AST
  • the warning message about Perl 5 regex modifiers was overzealous
  • it interpreted letters with whitespace in between
  • if appeared like a case-insensitive match
  • tightened up the error checker
  • added to my list of core types
  • thinking very hard about coercion and not so hard about trait auxiliaries

Patrick:

  • busy and productive week
  • worked on lots of little bug fixes
  • fixed .get and .lines to do the right thing: no extra lines, Nil when necessary
  • refactored postcircumfix brackets for the Positional role
  • it can do things in terms of itself
  • if you create a new class, you only need to define postcircumfix int
  • the others can define themselves in terms of it
  • you can define the others if you want, but you don't have to
  • had to update the handling of Whatever to return a closure
  • we were doing that for all operators except minus
  • that depended on getting postcircumfix fixed
  • ended up creating a new WhateverCode class
  • operators involving a Whatever return a WhateverCode
  • we can multidispatch and do special things on that
  • don't know if it'll last permanently, but it works for now
  • refactored Rakudo's assignment to define it in terms of .STORE
  • brings us a step closer to correct vivification of arrays and array elements
  • can write proxy objects which do the right thing
  • lots more discussion about various speed improvements in Parrot, especially with respect to IO
  • working on HLLCompiler
  • today working on making Rakudo work from an installed Parrot
  • Parrot does not make that simple
  • Vienna.pm has agreed to fund me for one Rakudo day per week in 2009, similar to what they have been doing for Jonathan

Jerry:

  • preparing for the Parrot Virtual Machine Workshop
  • will have more time to devote to Rakudo and Parrot during and after YAPC

Will:

  • nothing Perl 6ish
  • trying to encourage Parrot committers to blog their reports and not just put themon IRC
  • trying to write more regularly about Partcl
  • it's slowly returning to the peak passing rate
  • that's dropped quite a bit; been about six months since I've been able to do this on a regular basis
  • lots of memory limitations
  • I've been trying to write small test cases that demonstrate memory leaks
  • chromatic has fixed a lot of those
  • we've reclaimed a lot of tests since last week
  • tracked down a mismatch in Parrot and Tcl Unicode handling
  • have a workaround in place

c:

  • fixing memory leaks in Parrot, Rakudo, and Tcl
  • there's a big one specific to a Rakudo dynop
  • the other big one is Parrot_Context recycling in Parrot
  • I have a plan, but it's tedious to get there
  • also editing the Parrot book

Patrick:

  • Larry, what are you sending to the regexp engine on IRC?

Larry:

  • snowmen and comets
  • you can use them as delimiters... I don't know

Patrick:

  • I don't know why it works

Larry:

  • if you start it with a snowman, you should have to end it with a snowman

Patrick:

  • I'll look into it
  • I suspect it's a PGE parsing bug
  • it gets to the end and finishes without requiring a closing delimiter
  • looking at the Parrot VM workshop page
  • three hours is a long introduction to Parrot and PIR

Jerry:

  • that's easy to adjust

Patrick:

  • an hour for an intro to Parrot is fine
  • my experience in teaching new languages and systems is that you can go into the history and abstract sense of how it exists
  • but you end up losing more people than you gain
  • it's better to get into code quick

Jerry:

  • I haven't developed slides yet
  • threw it out there as a strawman
  • the people who have slides (slides can I borrow) should speak up
  • who wants to talk and who can do it?

Patrick:

  • will there be names by these talks?

Jerry:

  • there should be
  • I can add slots for names

Patrick:

  • I'll do PCT and the Rakudo stuff
  • unless someone's dying to do it, I can do the PIR stuff

Jerry:

  • Allison's done that in the past
  • I can take it over too
  • we've changed venue
  • we'll have two or three classrooms at CMU
  • not certain what our constraints are there on attendees and room size
  • we have 48 attendees signed up
  • Robert's frozen signups until we figure out venue issues
  • hopefully we'll open that up today

Patrick:

  • three attendees from our Perl 6 group in Dallas are going to YAPC for the Parrot VM Workshop
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
 Full
 Abbreviated
 Hidden
More | Login | Reply
Loading... please wait.