Parrotsketch Every Tuesday @ 18:30 UTC | IRC Log at irclog.perlgeek.de/parrotsketch/today
Set by moderator on 17 April 2009.
00:58 rin1024_ joined 02:57 eternaleye joined 03:27 japhb joined 04:26 eternaleye_ joined 04:29 eternaleye__ joined 09:35 ascent_ joined 10:09 bsdz joined 10:11 bsdz left 10:35 ascent_ joined 14:24 bsdz joined, bsdz left 14:34 gariani joined 14:37 danielgrafael joined 16:37 rdice joined 16:59 contingencyplan joined 17:08 PacoLinux joined 17:34 particle joined 17:41 Whiteknight joined 18:05 allison_ joined 18:11 fperrad joined 18:17 davidfetter joined 18:21 masak joined 18:29 Util joined 18:31 Coke joined, NotFound joined, barney joined
Coke Hello. 18:31
allison hi
fperrad hello 18:32
barney Hallo
NotFound hola
cotto hi
Util hi
Coke The three sharp taps of the gavel.... wrong meeting.
why don't we start with fperrad?
masak hi
Tene hi 18:33
18:33 moritz joined, chromatic joined
Coke francois? 18:33
fperrad * start a XML parser (SAX like) with PGE/NQP
PGE grammar allows to be very close to XML spec
* practise the release manager guide 18:34
18:34 Infinoid joined
fperrad * I start release just after #ps 18:34
EOR
18:34 darbelo_ joined
Coke Thanks for doing the release this month. 18:34
allison?
allison - Checked in the new book structure and edits to several chapters.
- Split out package specification files for Fedora, SuSE, and Mandriva (working with Tim Nelson, aka wayland).
- Continued working on the calling conventions, checked in the PCCMETHOD changes. Working my way through the old C calls to Parrot_runops_* and Parrot_run_meth_* 18:35
- Speaking about Parrot at a Linux user group in San Francisco tonight.
EOR
Coke Thanks. barney?
barney Pipp is working with SVN HEAD again
cotto++ for fixing PhpResource 18:36
Attended PHP TestFest in Munich
18:36 kj joined
barney Played with Issues on Github, looks nice 18:36
.eor
Coke Thanks. 18:37
chromatic?
chromatic Failed to merge in my header refactorings before the release; working on them now. Will merge in small steps after the release.
EOR
Coke danke. 18:38
cotto?
cotto Parrot:
* removed deprecated Ref and SharedRef PMCs (last week)
* finished removing UnionVal from PMCs (with Whiteknight++, also last week)
* closed a bunch of random RT tickets
GSoC:
* I'll be mentoring darbelo, who wants to write decnumber-based BigInt, BigNum and BigRat PMCs
* sent off an introductory letter about how to get plugged into Parrot
.end
Also, Whiteknight will be co-mentoring
Coke danke.
My report:
14:33 -!- chromatic [~chromatic@069-064-234-010.pdx.net] has joined #parrotsketch
... that's not my reprot.
- stationkeeping (tickets, code review, etc.)
- partcl still dead; never did have time to work on it before the 1.1
release. Am hoping that if rakudo works against an installed parrot,
I can too.
.EOR 18:39
Notfound?
NotFound * Closed some old tickets 18:40
* A minimal work on ecmascript
EOR
Coke gracias.
Util?
Util Last week: Tuit deficit; only worked on testing Parrot 1.1. 18:41
This week: More work on the book's PASM reference.
q1q
EOR
Coke particle?
tene? 18:42
Tene * More Web.pm work
* Have a pastebin running on rakudo
* Got HLL line numbers into uncaught exception messages, but it failed a bunch of tests that relied on the exact message
* Played with custom sub traits in rakudo... interesting possibilities there
* Plan to work on HLL compat in the next week
.KTHXBAI
Coke tewk_? 18:43
(thanks to everyone, btw, even if I don't say it every time. =-)
Whiteknight? 18:44
Whiteknight * Helped cotto++ finish removing UnionVal from PMCs
* Dealing with some JIT-related fallout from that
* Looking at some of the inheritance-related issues, hopefully get a few fixed post-1.1
* Some misc writing/rewriting in /docs/book/*
* Queue 1 Question
EOR
Coke PerlJam?
cotto q1q
Coke rdice? =-) 18:45
rdice Thanks for asking. 18:46
No report right now.
Coke let's see... I don't have an actual list of committers in front of me; anyone have a report that I didn't call on? 18:47
masak
Coke shoot.
masak * Participated in NPW, and the hackathon after that.
* It was a very enjoyable/productive workshop-and-hackathon.
* Submitted 18 rakudobugs since last Tuesday.
* Patched Rakudo, adding server-side socket support (mberends++, Tene++, cosimo++)
* Patched Rakudo, enabling double-coloned type names to appear in the setting
* Various other small janitor patches to Rakudo
* Started, during NPW, to discuss ideas for a module database in Rakudo
* Most pressing, I think, is the lack of a standard location for Perl 6 modules
* My queue of things-to-do is maxed out, but in a good way
.eor
q1q
moritz also has two lines to report
Coke shoot.
moritz * ususal Rakudo patching
* motivated people to do GSoC stuff
EOR
Infinoid
18:47 darbelo joined
Coke Alright. we have four questions queued; Before that, anyone else with a report? 18:48
Infinoid <--
Coke shoot.
Infinoid * Some patch monstering and ticket management stuff. Some cage cleaning, other minor random fixes.
* I've got some stuff for mingw and netbsd ready to go in just after the release.
* I'll be updating dalek again soon; if you have a language you want it to track, please make sure it's listed on the Languages page on the wiki.
1;
Coke kj was here but is gone; pmichaud isn't here.. 18:49
so, let's hit the questions... 18:50
Q1: Util?
Util When the user stack (poptop, etc) was removed from Parrot's design, did the design leaders intend anything in particular to replace its function, (like a plain Array object, perhaps)?
Also, is this intent documented anywhere?
Util wishes for a DEPRECATED_REASONS file when spelunking old PASM/PIR code.
Coke Util: post 1.0, there' a Deprecated page on the wiki.
allison Util: there's a local stack stored in each context
Coke The thought pre-1.0 was, for a while, if it's gone it's GONE, better to document what is. 18:51
allison (and, actually, the user stack is still there, it's just not rolled back with continuations anymore)
Util That all helps; I will dig deeper. EOQ.
Coke Q2: Whiteknight?
Whiteknight I propose we rip out the current JIT system. Make a clean start for the LLVM work 18:52
comments/opinions?
cotto That's my question too.
Infinoid I bet that would close a lot of tickets.
Coke Can't do that officially until 1.4, no?
Whiteknight is in ur mind, stealin ur questions
Coke er, until just after.
Whiteknight right, we we can start the train now
Coke but in principal... anything that gets me a JIT on my platform sooner++. =-) 18:53
allison let's see if the new LLVM JIT works before we start ripping out old stuf
moritz wouldn't it be better to wait with the rip-out until we *know* that the new approach works?
chromatic Ripping it out would hurt the JIT-built NCI signatures too.
Whiteknight not having a half-built JIT system will help motivate us to create a proper one later
cotto agree
Whiteknight because a lot of our target platforms don't have a working JIT now anyway
moritz also, does llvm satisfy our demands of portability?
cotto *agreed
allison the summer of code project will be done in a branch, and he can take any approach he needs, including entirely replacing it
Coke I think that doing it in a branch ... right.
NotFound Will be nice to be a nci call mechanism that does not depend on jit 18:54
Util parallel development, until new is proper superset of old.
allison but, jumping it to declare the existing JIT dead before we even know if we have a replacement is premature
Whiteknight I'm thinking more about motivations. We don't have a portable JIT system now, and the sooner we rip it out the sooner we can get started on making one
allison I appreciate your eagerness
chromatic I'd rather have something finished and usable before we remove something partially working but used. 18:55
Whiteknight people are less motivated to make a real JIT system when we have a phoney one already that sort of works when some people are looking
NotFound In theory it must not be so difficult to push the arguments on the stack with a little inline assembler.
allison it's not really *that* much difference to motivation
Infinoid The current jit has been intimidating enough to prevent me from seriously looking at adding platforms to it
Util People are motivated by shiny things. Build your shiny thing next to the old one; it will look even better.
Coke I tend to doubt that "working in trunk" is motivation for most people.
Infinoid I am interested in jitted arm
and x86-64 18:56
allison Infinoid: sure, that's a limitation in the current implementation
so, it certainly needs changes
and replacing it is one possibility
Util It does not even have to be in a branch. Make a JIT2 runcore.
Whiteknight I'm interested in x86-64 too, and I'm not going to take the herculean task of adding it to our curent JIT system. Too little return on investment
allison it's just too early to make the call
Whiteknight that's fair, I just wanted to sow the seeds of discontent. EOQ 18:57
Util (in trunk)
Coke ok. that's Q2 and Q3.
Q4: masak?
masak capturing output from a process: I wrote an email about it on parrot-dev the other day. it would be good for both Perl 6 and Parrot. is it on someone's roadmap and, if not, where do I start my own investigation?
Whiteknight masak, could we copy the backticks mechanism from Perl5?
Coke masak's post is lists.parrot.org/pipermail/parrot-d...02067.html for those playing at home.
masak Whiteknight: in what sense? is there such a mechanism for Perl5-on-Parrot already? 18:58
I'm all for copying the semantics from Perl 5, that's what I'm after.
allison we can't copy from Perl 5 because our I/O system is entirely different 18:59
all it needs is a full implementation of pipes in the I/O system
there's a partial one there now
masak ok, so that's the block.
allison needs a little work (on multiple platforms)
masak fine. just noting that it'll be good to have eventually. 19:00
19:00 pmichaud joined
Whiteknight let's get a list together soonish about all the platforms that still need work on that front 19:00
masak eoq
allison I can point anyone who's interested at the routines
Whiteknight <-- interested
allison Whiteknight: it's linux and windows
(Mac just uses the same as linux)
(but still needs separate testing) 19:01
pmichaud hello everyone
NotFound It needs some more that that. For example in windows is hard to make std redirections and creating a process avoiding the creation of a console window in all OS variants.
Coke and that's a wrap..... oh, hai. =-)
why don't we let pmichaud report first. =-)
allison NotFound: yes, that's windows
pmichaud ehhh, I wasn't planning to be here anyway. My flight is delayed in Newark.
I'm writing a report on our NPW results for use.perl and rakudo.org; I'll let that be my #ps report. 19:02
(probably post tomorrow morning)
Coke that works.
pmichaud eta to release?
Coke fperrad: anything you need done for the release?
fperrad no
all seems ok 19:03
Coke pmichaud: "just after #ps" was the last we heard.
so, shall we let fperrad get to work?
pmichaud okay. I have a patch I did on the plane that would be nice to get into the release (for rakudo)... but I think I'll err on the side of safety this month. 19:04
Coke danke.
pmichaud don't want to hold up the release. :-)
Coke our motto should be "there's always next month."
pmichaud well, it means that Rakudo will only be good against the current parrot release for a few hours :-(
Coke heh
barney next month never dies 19:05
Coke alrightee. That's a wrap.
see everyone after 1.1.0
19:05 chromatic left, pmichaud left, Util left, Coke left 19:06 masak left 19:07 fperrad left 19:08 NotFound left, darbelo left 19:41 allison left 19:44 Infinoid left 21:03 Whiteknight joined 23:28 particle1 joined 23:54 davidfetter left

Powered by Cro and the Raku® Programming Language.

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

Thank you!