Parrotsketch Every Tuesday @ 18:30 UTC | IRC Log at irclog.perlgeek.de/parrotsketch/today
Set by moderator on 8 December 2008.
14:18 kj joined
kj won't make it tonight, so preposting now: 14:22
==PIRC
+ bytecode generation works great
+ all basics work, non-keyed ops and sub calling
+ no fancy stuff: no flags except :named works (kinda)
+ there's some bugs somewhere in iterating the data structures
+ would like some carification on how keys are stored in bytecode
.end
bye
14:23 kj left 16:40 pmichaud joined, masak joined 17:31 Coke joined 17:32 mberends joined 17:41 Wknight8111 joined 18:01 rurban joined 18:25 barney joined 18:27 jhorwitz joined 18:29 chromatic joined 18:30 tewk joined
pmichaud good localtime 18:30
masak \\o
barney hi
chromatic hello
jhorwitz yo 18:31
chromatic Let's start. barney? 18:32
18:33 allison joined
barney [Pipp] 18:33
Started on support for closures.
Fiddling with objects.
Some cleanup in grammar.
[Parrot]
Applied some patches.
.eor
chromatic allison? 18:34
allison - Started, completed, and merged in a branch to make FileHandles subclassable/substitutable.
- Created a tasklist for GC work.
- Resolved some problems on the 'remove_pic' branch. 18:35
- Worked on integrating the changes from the calling_conventions branch into a fresh branch.
- Deleted old branch for PDD 30 'install' refactors. Reviewing fresh application of Reini's patches in fresh branch.
- Changed I/O modes so 'a' (append) also sets write mode.
EOR
chromatic I closed some bugs and fixed some compiler warnings.
I also sped up some STRING handling; in particular, PGE should be about a third faster.
Jonathan and I and others also discussed optimizations to parameter passing. 18:36
I have a very preliminary, dirty patch for pmichaud to benchmark PGE against.
I'll work on GC tomorrow.
Coke?
Coke partcl -
Trying to make partcl go faster and consume less memory.
In the absence of useful diagnostics, turns out this is hard.
Recent changes to partcl/parrot have lost several spec tests
(due to OOM PANICS)and quadrupled the run time of the suite.
parrot -
worked on deprecation a bit.
.
chromatic japhb? 18:37
japhb nr
chromatic jhorwitz?
jhorwitz fixed some minor mod_parrot bugs 18:38
found some issues wrt "use" semantics in rakudo, one of which is definitely a bug (RT #61826), the other was being discussed on the list (RT #61742).
with pdd22io_part3 merged in (allison++), i'm now working on ModParrotHandle PMC to serve as stdin/stdout in registry-style handlers. will release mod_parrot 0.5 when that's done and tested.
got a VMware ESX environment up and running for multi-platform testing :)
some foundation accounting tasks
q1q
EOR
chromatic masak? 18:39
masak * Been distracted from #ps lately due to end-of-term at uni.
* Created 63 tests this month.
* Other hobbies:
** Hacking on November the Perl 6 wiki.
** Hacking on a 3d board game called druid.
** Thinking about improving S29.
* Not a lot of Rakudo patching lately.
^D
chromatic particle?
rurban? 18:40
rurban - still not git, svn + quilt is better for me, so far.
- pdd30install_stage3: finished. Allison does the review. Then it can be merged.
make reallyinstall is then deprecated.
- did the cygwin release of 0.8.2, based on that branch. Finally found the blocker
for the previous release 0.8.1 which I missed.
- working now on 40817-MANIFEST-generated.patch,
with a small tools/build/addgenerate.pl
also in pdd30install_stage3 (not yet in)
- one or two tiny fixes in trunk
^D
chromatic Tene?
tewk? 18:41
tewk Working on class registry attempt two. EOR
chromatic Wknight8111? 18:43
Wknight8111 * Mostly did $reallife stuff and not much work on Parrot
* Followed some tickets around and talked on the list
* Some idle twiddling on the GC and some other things.
EOR
chromatic jhorwitz, you had a question.
pmichaud q1q 18:44
jhorwitz yes...
allison: i tried subclassing StringHandle in PIR -- didn't work out of the box. IIRC you said that should work? 18:45
allison jhorwitz: should, yes, but didn't write tests for that. let me know what problems you encountered
jhorwitz ok, we can take the details offline...
that's all for me 18:46
chromatic StringHandle is still using STRINGs incorrectly I believe, but I need to write code to demonstrate that.
pmichaud?
pmichaud q: can I give my report, now? ;-)
masak :) 18:47
chromatic Go ahead.
pmichaud ** Rakudo spectests (r34615): 264 files, 5912 passing, 0 failing
+79 passing tests since last #ps.
== Parrot stuff
: Had a long and productive discussion with allison++ about the class registry
: Here's the summary
. Using string class names such as $P0 = new "Foo" is highly discouraged
. Use $P0 = new ['Foo'] instead
. Class names in new/subclass/isa etc. are always relative to HLL root
. To use a class from another HLL namespace, one must do so via the
foreign namespace. This includes fetching 'parrot' classes.
. Internally, Parrot should use class objects instead of strings for classes
: In light of the above, PGE, PCT, and Rakudo will begin migrating
completely away from the $P0 = new "Foo" variants.
: Fixed some compiler warnings
== PGE stuff
: Now properly recognizes adverbs with leading spaces in regexes
: Added the "goal matching" syntax from S05 '(' ~ ')' <expression>
== PCT stuff
: Finished loop refactor, closing a Jan 2009 milestone (Tene++)
: New code is much simpler, straightforward
: In rvar branch 18:48
. added 'stmts' :pasttype
. new :scope('register') capabilities
. made .symbol() a bit smarter
== Rakudo stuff
: Added 'last' and 'redo' exceptions
: Adjusted infix:<~> and prefix:<~> to always force Rakudo Str objects
: Added Match objects
: Fixed .map/.grep/.values etc to treat scalars as lists of 1 element.
: Fixed parsing of pointy blocks in term position.
: Adjusted autothreading of infix:<~~>
: Working on a major refactor of variable and parameter handling, going well
. Array and Hash parameters now dtrt
. Create and attach explicit Signature objects to subs
. Now complains about variable redeclaration in same scope
EOR
chromatic Other questions?
Concerns about the roadmap? 18:50
Requests for help (besides partcl)?
Coke is GC actually going to hit for this release?
allison Coke: a GC refactor will
the incremental GC module won't 18:51
Coke ok. parrotRoadmap says "Gc refactor" but points to TT#3 which is "new gc system"
(setting expections)++
allison should change that to point to a different ticket 18:52
or, to the GC tasklist wiki page
Coke The deprecation milestone for 0.9.0 is not workable.
allison Coke: which? (looking up page now) 18:53
Coke (it says "remove all" - there's stuff in there that we just don't have the tuits for, like chip's old request)
pmichaud on the roadmap
iirc, at pds one of the goals for the 1.0 release was to do a release with no outstanding deprecations
allison Coke: oh, push that to the next month 18:54
Coke i don't see that happening, either, if we're going to keep to our schedule.
allison Coke: it's only critical for 1.0
Coke: what are the blockers?
chromatic Ugh, PMC union.
allison okay, that's not going away before 1.0
it may be that we need to remove some old deprecation items from the file 18:55
Coke so, we probably need to split up that "all deprecations" into chunks.
particle arrives, without news to report
allison Coke: yes
Coke which conveniently is already done in the .POD. =-)
pmichaud keep in mind that if we want 1.0 to have no outstanding deprecations (however we define it), then Feb 2009 is our last opportunity to deprecate anything.
(before 1.0)
particle let's concentrate on the user-facing deprecations 18:56
Coke I would recommend adding a second release to each deprecation; the release by which it MUST be removed. then PMC union can be marked as '1.5', or something.
particle namely, opcodes, exported functions, and pmcs
allison frankly, I'm fine with still having deprecation notices in, as long as they're attached to a sane future release, like 1.5 or 2.0
Coke: don't list 2 versions, just change the version number the deprecation is scheduled for 18:57
Coke also, I've done all I can in the remove_pic branch; I can only rip stuff out, not fix it. =-)
allison Coke: well, all tests are passing now
Coke: do you have language test problems?
Coke I haven't tested languages there.
Wknight8111 have you tested it with all the runcores? 18:58
allison Coke: or, problems with different runcores?
Coke I didn't even realize all the tests were passing.
allison Coke: I made a few fixes this week
Coke i had pretty much given up on it after the confusion about what the original ticket was asking.
I'll revisit.
allison the original ticket was pretty vague
Coke I have some more q's about deprecation stuff that can go to #parrot. 18:59
particle the whole ticket doesn't need to be addressed in order for the branch to merge
allison if it doesn't cause any new language failures, and doesn't cause any new runcore failures, I say merge it in
particle but fulltest does need to work everywhere
rurban On the roadmap there's my name at two critical points which I don't understand: ops tools, pmc tools. I have time, looks simple, but no idea what to do. Please add someone the tickets, email or irc log url. 19:00
allison rurban: which release?
rurban the next. 0.9.0 19:01
allison oh, for the pdd30-install task
rurban I put now a "?" to it.
pmc tools and ops tools are finished imho in my branch
I didn't touch them, besides the new 40817-MANIFEST-generated.patch which logs the filenames 19:03
Is it that what it should be? RT#40817. but that's not critical 19:04
allison rurban: will review RT #40817 after #ps 19:08
any more questions? 19:16
rurban nope 19:18
allison Then that's a wrap, thanks everybody! 19:19
19:21 masak left, Coke left, rurban left 19:23 pmichaud left 19:28 Wknight8111 left 19:37 mberends left
tewk :q 19:39
19:40 allison left 19:44 chromatic_away left 20:10 jhorwitz left 21:52 Whiteknight joined 22:34 davidfetter joined

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!