Priorities for this week: irclog.perlgeek.de/parrotsketch/201...#i_3798168 | Post closed tickets in your report. | Note: This channel is for our Tuesday status meetings (at 20:30 UTC); you probably want #parrot instead. | irclog: irclog.perlgeek.de/
Set by moderator on 24 May 2011.
01:00 cotto_work left 01:51 whiteknight left 03:17 lucian joined 03:27 lucian_ joined 03:30 lucian left 03:36 lucian__ joined 03:37 lucian_ left 03:40 lucian__ left 09:44 contingencyplan left 11:50 cotto left 12:49 bluescreen joined 13:37 contingencyplan joined 14:18 bluescreen left 15:21 cotto joined 15:27 bubaflub joined 15:59 cotto left 16:28 smash joined 16:58 cotto joined 17:26 smash left 17:40 cotto left 17:52 cotto joined 18:05 cotto left 18:26 Patterner joined 18:43 atrodo joined 18:51 cotto joined 19:06 kid51 joined
kid51 kid51's report 19:06
* DONE
** Posted to parrot-dev re request for status update on 3.6 Roadmap Goals.
** Closed no Trac tickets this week :-(
** Overall count (Trac report #10): 518; net decline of only 2 this week.
** Worked on YAPC talk; still need to do slides.
* WILL BE DOING
** Going to YAPC 19:07
** YAPC presentation, scheduled for Mon Jun 20.
EOR
tcurtis DID: 19:20
* Moved into summer apartment
* CFSM generation, but it's buggy
WILL DO:
* Blog
* Fix bugs
* LR(0) parser generation
EOR
19:28 mikehh joined
bubaflub DID: 19:29
* Parrot-GMP
- Major bug fixes; we're now segfault free
- Increasing test coverage
- Other minor improvements
WILL DO: 19:33
* Parrot-GMP
- Get test coverage to 100%
- Make setup.pir work 19:34
- Add metadata so we can install from plumage
EOR
19:40 NotFound joined 19:44 benabik joined 19:52 ligne joined
cotto #done: 19:53
- thinking about M0, found a hole in string processing (i.e. there isn't any)
- cut the 3.5.0 release without incident
- work on Parrot talk
- at Open Source Bridge, will give Parrot talk tomorrow (will be recorded)
- nudged soh_cah_toa toward a testing infrastructure
#to do:
- spec/implement some useful primitive string processing primitives
- give Parrot talk
- hack soh_cah_toa's test prototype into something more reusable
- implement M0 CPS factorial or fibonacci example
- think and write about what M0's future in Parrot will look like
- hang out with dukeleto++ and other hackers at OS Bridge and YAPC::NA
#blockers: 19:54
- way too many interesting talks and awesome people at OS Bridge
#eor
20:14 ligne left 20:17 ligne joined 20:19 whiteknight joined
NotFound What I did: 20:21
-parrot
* Minor fixes.
* Testing winxed snapshot.
-winxed
* Minor fixex.
* Minor tests improvements.
What I will do:
No plan
EOR
benabik DID:
* Added .annotate 'file' to PCT in master (TT #2132)
* Killed a few more Q:PIR blocks
* Caught up with bacek
* Looked over newPOST
WILL DO:
* Design PAST -> newPOST conversions
* YAPC::NA
* Stop worrying as much about removing every last Q:PIR
* Blog
BLOCKERS:
* My single-mindedness about killing PIR blocks
* Fixing my house
* Driving 16 hours to YAPC::NA
EOR
mikehh What I did since my last report: 20:22
* building and testing parrot on amd64/i386, with gcc/g++
* a couple of fixes
What I intend to do in the next week:
* testing and fixing
.eor
whiteknight WHAT I DID: 20:25
* Answering questions for GSoC and otherwise
* Talking lots about NameSpaces, lexicals, and other related stuff that parrot is doing wrong
* Doing what little I could to help dukeleto++ finish his grant
* Added vectorized testing to Rosella, need to blog about that. Also working on a big refactor of the Event library
* A few small cleanups to embedding/extending, although I quickly got in dukeleto's way. Will get back to that now that his grant is over.
WHAT I WILL DO:
* Get back to blogging
* Stare more at embedding/extending code. I have some stuff I want to write up about that
* Stare more at profiling code.
* Finish up some Rosella stuff
EOR
20:25 darbelo joined
Util # Done: RosettaCode, and worked on talk for YAPC::NA. 20:28
# Plan to do: Finish talk and attend YAPC::NA + Hackathon
# Blockers: $WORK
I will be in and out of the #ps meeting today.
.end
cotto hi 20:31
Util Hello
benabik ohai
bubaflub hello, q1q
whiteknight hello
darbelo Eclesia: Fun tidbit, at work naming a o/
tcurtis Hi.
darbelo wt
benabik bubaflub: already?
20:32 soh_cah_toa joined
kid51 Hello 20:32
cotto The release went smoothly, which was good because it meant that I could go to bed earlier.
bubaflub benabik: just queueing one question for later.
cotto What about other things that happened?
Or things that didn't.
kid51 Commit activity in master was sparse in the past few days, which I think only partially reflects pre-release status. 20:34
My casual impression is that most activity in past week was in GSOC or older branches
cotto kid51, I agree. I know that most of my efforts have gone toward M0 thinking and prototyping.
kid51 On parrot-dev, I posed question re where do we stand re July 3.6 roadmap goals. Have not gotten responses yet. 20:35
cotto I intended to reply but got busy with osbridge prep.
whiteknight I've been thinking about that
nothing to talk about yet, but I have been thinking 20:36
cotto Thinking is good. Blogging is also good.
kid51 The short version of those goals: benchmarking HLLs; plan better profiling; relationship managers with Rakudo
cotto The grant is off dukeleto's plate now. I won't extrapolate that he'll necessarily have more tuits. 20:38
Nothing's happened with profiling.
rather, I haven't done anything substantial there
kid51 whiteknight: Will you be able to begin focusing on profiling during the coming week? 20:39
whiteknight kid51: Yes, I looked at it before. I was going to wait until cotto freed up, but that's looking like an endless wait :)
so I'll stab ahead on it myself
kid51 whiteknight: Indeed! 20:40
cotto whiteknight, don't do that.
"Free" is a relative term in this context.
kid51 I think the 2 of you need to recruit some other people on the profiling effort.
cotto and working on both M0 and profiling is reasonable
kid51 s/other/additional/
cotto We need a plan first.
Right now I wouldn't have anything especially helpful to tell a willing hacker. 20:41
kid51 Plan was, by 3.6, to have "studied our existing profiling tools, determined their strengths and limitations and developed a plan for significant improvements in later supported releases."
whiteknight yeah, once I figure out where we stand and what is needed, I'll draft some fresh faces
cotto better imcc line numbers wouldn't hurt
soh_cah_toa :) 20:42
cotto It'd probably cure some of soh_cah_toa's nightmares.
NotFound Forgot to say Hola, sorry.
soh_cah_toa some, yes ;) 20:43
in all seriousness though, i'd love to see that
benabik cotto: Not just soh_cah_toa's.
cotto you're not alone. There's a reason we've had several attempts to fix it.
whiteknight If we can deprecate PIR heredocs, we can fix linenumbers pretty quickly, I think
cotto whiteknight, that'll mean rewriting a lot of tests. 20:44
benabik whiteknight: That's not the only issue, unless the heredoc parser screws up things when heredocs aren't used. IMCC is also wildly off the generated PIR for nqp_pct.
whiteknight IMCC has a lot of problems. Bending over backwards to make heredocs work is one aspect of it 20:45
cotto: I'm not saying it's a good course of action, just saying it's one possibility that will get us there
cotto I'd be working on it now but I've never gotten my head around lex and yacc
I don't think we'll resolve this issue today. I *really* want to get us some kind of pir compiler with sane line numbers, and imcc is the current "working" option. 20:49
whiteknight hopefully GSoC puts us in better position with respect to PIRATE 20:50
or, something similar
NotFound winxed does not generate pir heredocs, but line numbers sometimes are wrong anyway.
whiteknight NotFound: it's not the heredocs themselves that are the problem, it's the stupid decisions IMCC makes in order to accept heredocs
NotFound Agree
whiteknight line numbers are broken in part because of code that supports heredocs
tadzik I have a feeling that heredocs don't really fit in a thing like PIR, fwiw 20:51
whiteknight says nothing
cotto tadzik, that's not an inaccurate feeling
NotFound Sometimes I think the better solution will be: stop using lexx and yacc!
cotto NotFound, accepting patches
tadzik cotto: how commonly is it used in todays code? 20:52
cotto tadzik, it's very common in PIR tests
tadzik I see
benabik I should poke my head into IMCC. I actually somewhat like lex and yacc style parsers. (Admittedly, I usually use flex and bison.)
whiteknight IMCC is much better than it used to be, which is sad 20:53
cotto TimToady suggested that the correct way to do line numbers is to map the position in a file to the line number and look them up as needed rather than incrementing the line number at the right time.
NotFound That may be another solution. Using bison there are ways to automatically set position information into tokens. 20:54
whiteknight that's a fine approach, for our next compiler
cotto whiteknight, you don't think it's an option for imcc? 20:55
whiteknight as it currently stands? I don't know. We'd have to rearrange a lot of spiderwebs
and we don't have a lot of spider-bite insurance
cotto get torch 20:56
pmichaud mentioned that even basic sub-level profiling was very helpful in finding a hot spot. 20:57
A better imcc is highly desirable, but we can produce something useful with it as it stands. 20:58
whiteknight yes, we have to get that figured out too. I wonder if using a subclass of Sub would be a useful exercise, if we can make that work
or a subclass of CallContext
I still don't know enough about the system. I need some dedicated studytime with it 20:59
cotto It sounds like we at least know what the next step is. 21:00
whiteknight, do you mind if we move on to bubaflub's question? 21:01
whiteknight sure
cotto bubaflub, go ahead
bubaflub our docs for the NCI system is in PDD16 and it's still in the draft subdirectory. i was wondering what it would take to move it out from draft - i'm willing to help edit or bring it up to date. if it's too much work, i was wondering what kind of docs people would like to have for the NCI system. 21:02
cotto bubaflub, you're welcome to bring it up to date. 21:04
bubaflub cotto: certainly. would people like some kind of tutorial as well? i'm thinking something more narrative-esque
cotto If it's current and accurate, there's no need to keep it in drafts.
mikehh needs some work though
whiteknight should definitely involve plobsing, make sure his knowledge finds its way into the document 21:05
cotto +1
whiteknight also, since his is the driving vision in that subsystem, he could help to spell out the future plans for it
mikehh and I think tutorials are always helpful
bubaflub whiteknight: i concur; plobsing expressed mild disdain for PDDs however; he said he'd prefer to see a tutorial more than a PDD. 21:06
at the very least i'd like to bring it up to date - 't' deprecation, better examples, and so on
and i can run the edits by him before they land in master
cotto The system needs to be documented in a useful format. That's my biggest criteria.
whiteknight bubaflub: don't tell plobsing that you're working on the PDD. Ask him a bunch of questions and write down the answers in the PDD when he isn't looking 21:07
NotFound Certainly, writing a document supposed to drive the design following an existing design is less than idea.
ideal.
bubaflub whiteknight: sneaky sneaky
NotFound: agreed. but at least it can be current; i'm not exactly sure what the future plans are.
cotto It's closer to ideal than what we have now, so +1.
NotFound AFAIK the future plans are: "Improve it"
bubaflub ok, i think i will update the PDD whilst running the changes by plobsing surreptitiously and i'll also have a tutorial 21:08
cotto excellent
bubaflub the tutorial will contain things not specifically related to NCI like StructView, String, tests, etc.
ok, eoq.
NotFound BTW, q1q 21:09
cotto NotFound, go ahead
NotFound UnManagedStruct does not implement vtable get_bool, I think using to return true if not null will be an improvement. What do you think? 21:10
whiteknight NotFound: why?
(i'm not against it. I just want to know the motivation)
cotto What's the use case?
NotFound whiteknight: nci call that return a maybe null pointer. 21:11
whiteknight ah, okay
cotto That's a good use case.
+1
whiteknight has to go make dinner. Will backscroll 21:12
cotto any other comments or questions? 21:14
NotFound No opposition and just after a release is a good time to give a try. EOQ.
mikehh so the get_bool will return true if something exists? 21:15
cotto For those planning on coming to yapc::na, get ready in whatever form is appropriate
NotFound mikehh: if the UnManagedStruct has been assigned something.
mikehh NotFound: sounds good to me 21:16
NotFound Something other than NULL, that is.
cotto NotFound, yes, it's a good time to break things. 21:17
Does anyone have suggestions for weekly goals? 21:19
NotFound Break things!
kid51 1. First draft of plan for profiling (whiteknight)
2. Attendees at OpenSource Bridge should blog or post to parrot-dev about it
3. Same as 2, but for YAPC (which will be in progress at this time next week) 21:20
Attendees at YAPC should check in with me upon arrival in Nashville; #parrot msg will suffice. 21:21
s/Nash/Ashe/
cotto btw, I'm sitting next to the guy behind CoApp and am planning on using it to build a package for windows. I'll add that as a goal, though I won't be able to do anything until July when I have a windows machine at work. 21:22
any other goals? 21:23
GOAL 0: explore what's needed to produce a useful profiler (cotto, whiteknight)
GOAL 1: break things (and fix them) (NotFound)
GOAL 2: OS Bridge blogging
GOAL 3: YAPC::NA blogging
GOAL 4: CoApp package (cotto)
coapp.org/ 21:24
21:25 kid51 left
NotFound Secondary goal: think about some application relatively light that can benefit an embedded parrot. 21:25
cotto xchat and the GIMP were mentioned previously
NotFound Lighter than gimp, preferably.
cotto Let's call it a wrap. 21:27
21:28 bubaflub left
ligne one last thing. i brought this up in #parrot, but bubaflub suggested i mention it here too, to get some feedback... 21:28
i've had a little play at adding .pir support to checkdepend.t: <github.com/ligne/parrot/commits/checkdepend> 21:29
NotFound ligne: I think you'll get more feedback with an msg to parrot-dev 21:31
ligne NotFound> fair enough 21:33
21:39 cotto left 21:46 Patterner left, PurityControl joined, PurityControl is now known as Patterner 21:51 NotFound left, soh_cah_toa left 22:17 benabik left 22:20 particle1 joined 22:23 particle left 22:29 cotto joined 23:11 Tene left 23:19 whiteknight left 23:36 whiteknight joined 23:54 Tene joined 23:58 cotto 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!