Priorities for this week: Pre-post/plan PDS topics & attend PDS; Damn the build, full Awesome ahead! | Post closed tickets in your report. | Note: This channel is for our weekly status meetings (Tuesdays at 19:30 UTC); you probably want #parrot instead. | irclog: irclog.perlgeek.de/
Set by moderator on 13 August 2011.
00:55 darbelo joined 03:48 cotto joined 03:49 darbelo joined 06:41 cotto joined 08:44 cotto joined 09:30 Tene joined 09:41 lucian joined 10:09 cotto joined 12:42 bluescreen joined 12:49 darbelo joined 12:59 cotto joined 13:10 Tene joined 13:19 darbelo joined 14:37 darbelo joined 15:34 Tene joined 15:37 darbelo joined 16:59 bubaflub joined 17:58 contingencyplan joined 18:32 Coke joined 18:46 NotFound joined
Coke Doing: cutting release today. Unpleased to find 'make release' borked on OS X, but will avoid the issue for this release. Expect a cut in the evening, Eastern time. 19:05
.
19:05 atrodo joined
NotFound What I did: 19:05
-parrot
* Updated winxed snapshot
-winxed
* Added :multi capabilities from whiteknight++ fork
* cast to var builtin
* Set release number to 1.1.0 for parrot 3.7
What I will do:
No plan
EOR
lucian what i did: 19:08
fixed many bugs in the object system
added functions in the compiler
54 tests pass, 6 fail
what i will do:
blog
write docs about the object system
EOR
bubaflub what i did: 19:09
not much - real life interrupted
what i will do
finish docs
blog
EOR
19:15 soh_cah_toa joined
soh_cah_toa REPORT 19:17
* Wrote a few tests
* Started designed spec for SOD format
* Blogged
TODO
* Final blog (yikes!)
* Try to fix test suite to run more than one command
* Get people working on SOD design
EOR
q1q 19:18
Util # Done: Nil. # Plan: $WORK. # 7-day ticket report: 5 closed, 6 new. .end 19:23
19:25 cotto joined
cotto #done: 19:31
- not_gerd poked some important holes in M0
- not unfixable, but will require different philosophy
- at yapc::eu, hanging out with Rakudo folks
- removed parrot_debugger from install
- segfaults on --help
- --help is not the kind of thing we want to discourage
#to do:
- learn myself some Rakudo 19:32
- synthesize not_gerd's thoughts into a new direction for M0
- spell out what needs to change and what won't
- give parrot state talk at yapc::eu
- finish profiling doc
#blockers
- too many interesting things at yapc
#eor
also, hello
19:32 kid51 joined
soh_cah_toa howdy 19:32
tcurtis hello
kid51 hello
tcurtis arrived too late to pre-report. Will report after the meeting. 19:33
cotto tcurtis, go ahead. It'll be fun.
atrodo howdy
kid51 worked on PaFo business (membership roles); trac.parrot.org/parrot/ticket/2161, which should close after release; began sketching proposal for benchmarking 19:34
cotto How's this week been? 19:35
NotFound Hola
soh_cah_toa well, i started designing the debug format
Util Hello
soh_cah_toa which i'll bring up a little later
cotto soh_cah_toa, excellent! What tools are you using, and are the amenable to collaborative editing? 19:36
Coke kid51: when you cut the last release, I assume you did so on linux?
soh_cah_toa cotto: of course, it's in a gist
cotto: the link is in my blog
kid51 Coke: Yes (my mac is too old/slow for things like 'make world' :-( )
cotto needs to backscroll
Coke kid51: 'sfine. wouldn't have worked anyway. ;)
tcurtis Did: got LR(0) stuff actually parsing and building a parse tree, as well as performing actions. Wrote some tests for same. Started on SLR(k). 19:37
Will do: Finish SLR(k). Tests. Docs. Blog. EOR
cotto is there anything that needs attention before the release? 19:39
kid51 We have +1 more open TT ticket this week than last (Report #10): 514 total; we can't seem to get close to 500.
soh_cah_toa i have the soh-cah-toa/tt-2155 branch for tt #2155 which i can definitely merge and close this week
kid51 cotto: The Archive::Zip, Archive::Tar tickets are still unresolved -- but I don't think we have to correct them before this release.
I don't know of any barriers to today's release. 19:40
cotto I like the sound of that.
anything else before we move to questions? 19:41
19:42 benabik joined
cotto alright. soh_cah_toa, go ahead with your question. 19:42
soh_cah_toa sure
i'd like to bring up my idea for a debug format: "symbolic opcode description" format (sod) 19:43
i've pretty much explained everything about why we need this in my recent message to parrot-dev and in my blog but i want to mention it here because i'd like to see it recognized as a roadmap goal
and also i'd like to establish a team of people to get working on it
i think the lack of a debug format is one of the things holding us back from having real analysis tools like debuggers, profilers, instrumentation, etc. this can open up *a lot* of opportunities for us
so...any comments/volunteers? :)
19:44 whiteknight joined
kid51 soh_cah_toa: 1. How much existing code has to be rewritten to implement a SOD format? 19:44
whiteknight kid51: not a whole heck of a lot, because we don't have much here 19:45
soh_cah_toa kid51: i'm not sure anything really has to be rewritten though i could be wrong. it would mostly be adding to pct
benabik Do we already have a debug segment/format?
whiteknight benabik: not really
soh_cah_toa segment, yes. format, no
whiteknight we have a simple opcode->line_number mapping
soh_cah_toa and the segment is very poor
kid51 is there a particular meaning attached to "segment" here?
whiteknight it's a part of a packfile
cotto piece of a pbc file 19:46
benabik what they said
cotto or packfile struct, more precisely
soh_cah_toa docs/parrotbyte.pod explains it
kid51 q1q 19:47
soh_cah_toa so i'd need people w/ a good knowledge of pct 19:48
kid51 What is the difference between a debug segment and a debug format?
soh_cah_toa well, a format just refers to the format the segment is in
like dwarf, stabs, coff etc
benabik Basically I was wondering "do we have a place in our pack files to put this" and "is there something already there"? 19:49
whiteknight the format is the design, segment the implementation
soh_cah_toa whiteknight: yes, exactly
benabik: see my blog post or message to parrot-dev. it explains both
it's pretty much for storing high-level info about the original source in the bytecode 19:50
benabik I could help add it to PCT. We'd probably also need a new PMC to access/create the data.
kid51 Well, I applaud the idea that this is acknowledged as needing a *team* effort, and applaud soh_cah_toa for actually wanting to lead a team!
soh_cah_toa thanks! i think this could really open up a lot of new doors for us 19:51
and would also help the difficulties we've been facing w/ profiling
benabik Hard to profile accurately if you don't know what things are. 19:52
whiteknight since profiling is high on our list, if this really does turn out to be a stepping stone towards that, this moves up to the top of the list
soh_cah_toa my thoughts exactly
benabik Not sure it's a requirement for profiling… But would definitely help make profiling output more useful to HLL users. 19:53
soh_cah_toa of course
that's why i'm bringing it up. it's not just something i want, it's something that we (as in parrot) needs 19:54
cotto soh_cah_toa, a large part of the problem with hbdb was that you couldn't get decent line numbers from imcc. Is that something that you plan on addressing directly with this effort?
soh_cah_toa cotto: yup, i have a solution in the spec
cotto (apologies if that's covered in the p-d post or blog, which I'm reading as we go)
soh_cah_toa w/ a line number table
it's not. probably should've mentioned that :\\ 19:55
benabik IMCC's line number problem is due to the opcode/line mapping being insufficient?
NotFound IMCC's problems are... just that.
soh_cah_toa :) 19:56
benabik I was under the impression it literally had problems tracking the line number.
cotto benabik, I believe it's due to the implementation
soh_cah_toa there's a lot of different reasons and i think the (dwarf-borrowed) idea of a line number and address table solves that in a snap
cotto I'll look forward to reading your proposals. 19:57
soh_cah_toa great :)
cotto over the next few minutes ;)
soh_cah_toa so atm, i'm done picking out the things i like from dwarf
i just need help seeing how this can be implemented 19:58
and more pct-ized
lucian benabik: it does, if you mis-indent your PIR
soh_cah_toa b/c pct is an area i'm only just becoming familiar w/
cotto soh_cah_toa, I like where this is going. 20:00
any other comments or should we move to kid51's question?
soh_cah_toa yes, me too. i'm very excited
we can move on 20:01
cotto kid51, go ahead
kid51 Thx. soh_cah_toa: I've been following your hbdb branch 20:02
But have always gotten errors in the tests.
soh_cah_toa kid51: yes, and i thank you for that :)
oh yeah, that whole thing :(
kid51 I have reproduced those errors on 2 different Linux boxes.
But IIRC you have not been able to reproduce them.
soh_cah_toa no, it's so strange
kid51 Now, I did attempt to use hbdb natively ...
cotto and they look fine for me too 20:03
kid51 ... and, regardless of the IMCC problems, it seemed to DTRT as far as the basic commands go.
So I'm inclined to believe this is an iatrogenic error, i.e., a limitation in the testing apparatus itself.
soh_cah_toa kid51: indeed
kid51 I want to know what your plans are for that branch in the short run.
soh_cah_toa well, the first thing is to try to get tests that can run more than one command at a time 20:04
kid51 We know, e.g., that because it was created a long time ago, master has diverged in several respects, e.g., bytecode format.
soh_cah_toa yes
b/c atm, one test = one command
whiteknight a merge shouldn't be too too painful
soh_cah_toa no but all tests should pass 20:05
kid51 What I would like to do is to have other people, who know more about what debuggers ought to do than I do, review hbdb apart from the IMCC problems.
soh_cah_toa good idea
kid51 It may be time to refresh the branch from master
Or to fork a new branch from master and bring in the HBDB parts.
soh_cah_toa i think i tried that a while ago but you're right. this could help solve those problems w/ the tests 20:06
kid51 Writing tests for a debugger is, we must concede, a formidable problem in itself, because you have to be thinking in terms of both your program's STDOUT/IN/ERR and the debugger's as well. 20:07
soh_cah_toa yes, ipc programming/testing itself is a nightmare
kid51 So, to avoid getting wrapped up in testing conundra, we should have other people just try to use it and get feedback about that experience.
soh_cah_toa sure 20:08
NotFound Parrot std out and err can be redirected.
soh_cah_toa the next thing i gotta do is watchpoints
kid51 NotFound: If you can take a look at lib/Parrot/Test/Hbdb.pm in his branch, you might spot something I haven't. 20:09
(or whatever soh_cah_toa's testing module is called)
NotFound I'll try 20:10
kid51 EOMQ 20:13
cotto any other questions?
kid51 I have an announcement. 20:15
cotto announce away!
kid51 I'm slowly preparing an annual Parrot Foundation members meeting. 20:16
This entails a lot of inspection of commit logs, CREDITS, our Dropbox, etc., to see who qualifies.
So, this past weekend, I identified people who had submitted CLAs which were found in our Dropbox and who were not yet on the parrot-members mailing list. 20:17
That list probably had not been updated since last year's meeting. (Of course, there was no urgent need for it to be.)
I then subscribed all those qualifying individuals to the list. (If they want to be unsubscribed, they can simply follow typical Mailman link.) 20:18
That leaves several groups of people.
(i) People who meet the "made two contributions to Parrot" qualification but for whom I could not find CLAs in our Dropbox. 20:19
(ii) People who might have signed CLAs in the past (including in our Perl Foundation days) that I don't see.
So, this past weekend, I tried to send an email to those people urging them to sign CLAs, after which we could subscribe them to parrot-members, and after which they would be qualfiied to vote at annual meeting. 20:20
I'm now sure whether all those emails got to the people intended.
And some people may have responded ... but I'm not sure whether 'legal@parrot.org' is properly redirecting to 'parrot-directors@parrot.org'. 20:21
So, we have technical difficulties. And I'm the only one working on this.
So I can't say that we're ready to schedule an annual meeting for Aug 30, as I had originally hoped.
If you *have* tried to send a signed CLA in recent days, please ping me about that off channel so I can follow up with you. Thanks. 20:22
kid51 also requests that the other Board members please respond to my posts to parrot-directors. 20:23
kid51 also requests that the other Board members please return my phone calls.
whiteknight did I get a phone call?
kid51 No, but I've left 2 message for one of the others. 20:24
end_of_announcement
cotto kid51, thanks for doing the legwork to make the elections possible 20:25
would anyone like to make any other points? 20:28
kid51 How's YAPC::EU going?
benabik I'd just like to note that the channel topic appears to be empty.
cotto benabik, I see it
Util benabik: try saying /topic 20:29
cotto kid51, quite well. Rakudo is very well-represented.
I'm spending some instructive time with jnthn, masak and pmichaud.
I wish more Parrot hackers were here, but I can't make that happen. ;) 20:30
benabik Still not seeing it, but I'll accept it's a client issue.
cotto my client does that too sometimes 20:31
let's call it a wrap 20:32
benabik Ah. I can get it out of the server window… So I will instead note that the channel topic still talks about PDS. 20:33
soh_cah_toa sure, good meeting
Coke expect a release this evening, eastern time.
cotto expects
20:33 soh_cah_toa left 20:38 bluescreen_ joined 20:54 atrodo left 21:39 NotFound left 22:24 darbelo 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!