Priorities for this week: irclog.perlgeek.de/parrotsketch/201...#i_3126985 | 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/parrotsketch/today
Set by moderator on 20 February 2011.
00:14 whiteknight joined 01:19 cotto left 01:48 cotto joined 02:04 cotto left 02:22 tadzik left 02:23 tadzik joined 02:25 whiteknight left 04:21 cotto joined 05:46 bacek left 08:17 contingencyplan left 09:49 bacek joined 10:00 bacek left 10:12 bacek joined 10:45 lucian joined 11:37 lucian_ joined 11:41 lucian left 11:54 lucian joined 11:57 kid51 joined 11:58 lucian_ left
kid51 kid51's report 12:02
* DONE
** Parrot Foundation legal/financial business 12:03
** We're approaching halfway point between Parrot Developer Summits. Emailed people who named themselves as members of teams working on Roadmap Goals to assess progress.
** Cage-cleaning: reviewed about a dozen older tickets; this led to my or others closing about 6 tickets.
** trac.parrot.org/parrot/ticket/1049: solved 'make -C' substituted for 'make' problem.
* WORKING ON/WILL DO
** trac.parrot.org/parrot/ticket/1159: make release_check target
EOR
12:04 lucian_ joined 12:09 lucian__ joined, lucian left 12:12 lucian joined, lucian_ left 12:15 lucian_ joined, lucian__ left 12:17 lucian__ joined 12:18 lucian left 12:21 lucian_ left 12:23 lucian__ left 12:42 lucian joined 12:53 lucian_ joined 12:57 lucian left 13:08 lucian joined, lucian_ left 13:10 lucian_ joined 13:14 lucian left 13:23 kid51 left 13:38 wagle left 13:39 wagle joined 13:43 lucian joined 13:47 lucian_ left 14:44 lucian left 14:47 lucian joined 15:13 PacoLinux joined 15:22 PacoLinux left 15:24 contingencyplan joined 15:26 PerlJam left 15:27 PerlJam joined 16:27 atrodo joined 16:31 lucian left 16:33 lucian joined 16:39 lucian left 16:44 lucian joined 16:45 lucian_ joined 16:50 lucian left 16:51 lucian_ left 16:52 lucian joined 16:55 lucian_ joined 16:58 lucian left 17:02 lucian joined 17:05 lucian_ left 17:07 lucian_ joined 17:10 lucian left 17:16 lucian joined 17:19 lucian_ left 17:42 lucian left, lucian joined 17:44 lucian_ joined 17:48 lucian left 17:59 plobsing joined 18:06 lucian joined 18:09 lucian_ left 18:20 lucian_ joined 18:23 lucian left 18:26 lucian_ left 18:27 lucian_ joined 18:29 dukeleto joined
dukeleto won't be able to make #ps today 18:29
dukeleto wrote leto.net/dukeleto.pl/2011/02/parrot...agons.html recently 18:30
and yes, there are some small mistakes (like the length vtable isn't called that) that need to be fixed in that post
18:45 lucian joined 18:48 lucian__ joined
plobsing will also likely miss #ps 18:48
18:48 lucian_ left
plobsing What I Did: 18:49
* convert '.param' PIR syntax to a statement (no longer constrained to be at begininig of function). useful with coroutines.
* complete ptrbuf merge
8 * added probes for 'long long' and 64-bit C types
* added support for 'long double', 'long long', and 64-bit to StructView
What I Plan:
* revisit Inf/NaN work 18:50
* deprecate old pointer types
* begin nci improvements mirroring ptrbuf changes
* get back on the imcc improvement train
EOR
18:51 lucian left 18:55 lucian joined 18:56 NotFound joined 18:58 lucian__ left 19:14 ShaneC joined
NotFound What I did: 19:15
-parrot
* Testing, added one test
-winxed
* operator exists
* operator class
* operator .* (indirect method call)
* $loadlib directive 19:16
* Added -o option to installable driver
* Minor fixes and refactors
What I will do:
Test and improve winxed recently added features.
EOR
tadzik am I late? 19:21
seems no, cool
cotto_work tadzik: it starts in 57m 19:30
tadzik good, thanks
tewk q1q: What do people thing about merging tewk/select? 19:57
20:13 whiteknight joined 20:15 kid51 joined
whiteknight WHAT I DID: 20:15
- Hacking on the IMCC branch. Identified what I hope is the last major bug. Working on a fix.
- Translating Rosella into Winxed, providing feedback to NotFound++ in the process. About 50% complete.
WHAT I WILL DO:
- Continue hacking on IMCC. hope to have a fix ready for wide testing by this weekend. 20:16
- Continue work on Rosella. Would like to have it translated completely into Winxed soon.
- Getting excited about GSoC. Starting to brainstorm project ideas.
EOR
cotto_work *did: 20:19
- threw my brain at opsc, "helped" bacek get it to parse ops' C instead of just mangling it with regexes
- code review, feedback
- M0 roadmap progress
-- started digging into ffi
*will do:
- figure out a sane set of primitives for M0 ffi
- prioritize remaining open questions in M0 spec
- play with bacek++'s new opsc 20:20
*blockers:
- none
*eor
q1q
mikehh What I did since my last report: 20:21
* building and testing parrot on amd64/i386, with gcc/g++
* some fixes
* branch testing and fixes
* adding missing documentation to pmc's and
* removing passing tests from TODO list
What I intend to do in the next week:
* testing and fixing
.eor
Util # Done: 20:22
* Fixed 8 Coverity defects
* Fixed 2 Headerizer warnings. (one thorny warning remains)
* Misc other cage work.
# Plan to do:
* Fix TODO in t/op/exit.t
* Finish Rakudo Star binary dist for OS X.
# Blockers:
* None
.end
atrodo .did 20:24
* small amount of lorito hacking
* some futzing with isparrotfastyet, not visible changes
.todo
* more lorito hacking
* bite the bullet and rerun old commits with new tests with isparrotfastyet
.blocker
* yaks
.end
q1q
cotto_work hio 20:30
NotFound Hola
kid51 hello
mikehh hi there
bacek aloha
cotto_work How'd we do this week? I don't think we had any specific goals, but general comments are welcome. 20:31
we're down to 507 tickets, which is nice
Util q1q 20:32
cotto_work We should make it a goal to get some gsoc ideas on the wiki. It's coming up. 20:33
atrodo howdy
Util 7-day ticket report:
10 closed: fixed
1 closed: invalid
6 closed: wontfix
1 closed: worksforme
5 new
cotto_work If there are no comments on this week's goings on, we can move to questions. 20:34
mikehh is there any progress on smolder? 20:35
kid51 I have received no response at all from OSL on the bug ticket I filed
cotto_work none that I'm aware of
kid51: if it just needs to be restarted, they're pretty responsive in #osuosl on freenode 20:36
kid51 I don't have the knowledge to say whether a simple restart is needed or something else.
This is actually the first time they have not been responsive in bug ticket.
I will try again 20:37
cotto_work kid51: thanks
tewk had a question queued:
<tewk> q1q: What do people thing about merging tewk/select?
NotFound +1
kid51 What does it do?
cotto_work NotFound: have you reviewed the code at all?
whiteknight I haven't given it a close read-through, but it looks good to me
NotFound cotto_work: no, but given that is not subject to deprecation or something, it will be better tested in trunk. 20:38
I mean, master
cotto_work github.com/parrot/parrot/compare/m...k%2Fselect
kid51 Assuming that the code is well-written, (a) do we want to include its functionality in Parrot; (b) is it thoroughly tested? 20:39
cotto_work tested, yes 20:40
thoroughly, not sure
kid51 I ask because this was (like one other TT this week) a bit out of the blue
cotto_work the functionality is something portable and generally useful, so including it in parrot is welcome
NotFound The functionality is highgly convenient given out lack of good thread abilities. 20:41
bacek +1 to merge it.
cotto_work wfm 20:42
tewk: merge when ready
my question:
tewk Ok I might add a few more tests, but I will merge it soon.
cotto_work bacek and I have been working on making opsc smart enough to fully parse the C used to define our ops. It seems to work for core and Rakudo, but it could use testing and review. Are there any objections to merging it before the end of the week?
mikehh looks good to me 20:43
NotFound Provided that is not intolerably slow, +1
cotto_work (when I say "bacek and I", it basically means that I was able to sneak something in while bacek was sleeping and was therefore unable to beat me to it)
there's about a 20% speed hit, but that only comes into effect when rebuilding ops 20:44
NotFound Not so bad.
whiteknight those op-building tests are already unbearably slow
kid51 It built and tested fine for me in a basic Linux configuration
NotFound Given what I does, amazing.
mikehh will do some more test runs on it
kid51 cotto_work: Can you explain more about the speed hit?
cotto_work kid51: what do you mean?
kid51 Given that Rakudo et al are demanding more speed out of Parrot, it seems reasonable to ask why we should be slowing it down 20:45
cotto_work it's 20% slower in generating src/ops/core_ops.c, which is checked in
the generated code is essentially the same
kid51 Are you saying that this speed only affects Parrot 'make'?
cotto_work kid51: it can only slow down the build. Rakudo's runtime performance won't be effected. 20:46
kid51 If so, and if the generated code is essentially the same, what is the advantage?
tewk What about parrot or opsc makes it slower than the perl version of opsc?
Because it runs on top of parrot now.
cotto_work kid51: the advantage is that opsc has a semantic understanding of the C code now and can be made to output something other than C. 20:47
NotFound It parses the code instead of just figuring where the C code starts and ends.
cotto_work (or grammatical, at least)
NotFound I hope it will also hep to diagnose errors better and earlier. 20:48
cotto_work Our dream of converting the .ops files to lolcode is one step closer.
NotFound help
kid51 okay, while not objecting to merging it, I would really welcome something like a blog post that explains to lay people like me what it does and why it's better
NotFound +1 on that
cotto_work (or nightmare)
eoq then
kid51 We need to give people, like myself, who are not "deep C coders" something to point to when we talk about Parrot with others.
mikehh it's more self-contained and bootstraps 20:49
cotto_work mikehh: it did that before. Now it's smarter.
atrodo: go ahead with your question 20:50
kid51 If there were someone who could write a blog post or post to parrot-dev about that, we could point to that in our next release announcement.
mikehh yes and allows more options
atrodo I'm in the middle of working on isparrotfastyet, and dukeleto's "smoke-me" branches made me think about doing something similar 20:51
So, the question is, I could do composed graphs that have multiple tests on one graph (like the perl6-bench set) or multiple branches. Which do people think is more useful? 20:52
cotto_work I don't like the idea of putting metadata in a branch name.
whiteknight ditto
cotto_work atrodo: I'd prefer one graph with multiple branches. 20:54
atrodo That's what I was leaning towards. But that brings up the "which branches" question 20:55
Util You could use a smoke-me/ dir in a branch, containing a file of metadata. Just remember to remove the metadata file before merging.
cotto_work atrodo: can you make an interface in ipfy to add/remove branches? 20:56
kid51 At any given time, we only need to smoke master and 2-3 other branches that are being actively worked on. 20:57
whiteknight a metadata file in a metadata repo somewhere would be nice
kid51 But which 2 or 3? That changes week-to-week
whiteknight dalek reads from a .json file in perl6/mu. We could have the same kind of thing
parrot/meta on github
tewk I think the branch name is easy, ingenious, and convenient. branches are short-lived, parrot doesn't have to overcomplicate and over abstract every step in the development process. 20:58
whiteknight not *every* step
atrodo cotto_work> Adding an interface wouldn't be hard, i just worry about administration of who has access
whiteknight branch names are fine in theory. But what if we want a branch which is both smoked and benchmarked? smoke-me/benchmark-me/foo? 20:59
cotto_work My preference is a web-based interface, but the guy who's writing the code gets the most votes. ;]
atrodo Something like branch names or parrot/meta would levrage permission sets that already exists
whiteknight that's why I suggest the new repo. It could serve other purposes as well
cotto_work atrodo: a hard-coded password would be fine for the first approximation
(not long-term, of course)
parrot/meta sounds fine too. 21:00
atrodo Any objections to parrot/meta? Sounds like the path of easiest maintenance to me 21:01
cotto_work atrodo: go for it
atrodo wfm, eoq
cotto_work Util: go ahead 21:02
Util What was the original use of `attribute_defs' (in _vtable)? Does anyone remember?
I ask because I think it is unused, and am exploring removing it.
whiteknight i have no idea. 21:03
NotFound Util: looks like pmcproxy uses it 21:04
Util NotFound: Yes, only to copy/proxy it. It then gets used by nothing.
NotFound Ah, yes, is used to create the object attributes corresponding to the ones in a pmc parent. 21:05
cotto_work Util: I don't know without some archeology.
NotFound I've just remembered that I wrote it X-) 21:06
cotto_work NotFound: I love that feeling.
whiteknight so we want to keep it?
Util If anyone remembers a reason that I should not waste my time, just tell aloha. Otherwise, I will push a branch that removes it, for review.
NotFound I think that object inheritance from PMCs depend on it. 21:07
whiteknight if that's true, we should see that code in the coverage reports
if it's covered, we're using it. If not, we aren't
Util NotFound: Great! then it should break tests on removal?
whiteknight because we do plenty of inheritance in the suite
NotFound Util: I hope so 21:08
whiteknight: that doesn't prove much, the attributes will be built even if never gets used.
cotto_work The code in pmcproxy that deals with attribute_defs gets hit by our test suite. If it goes away and nothing breaks, there's a good chance it's safe to take out. 21:09
whiteknight so if the attributes get built anyway, why do we need that vtable?
cotto_work trying now
NotFound whiteknight: the attributes are built using the data in that vtable field.
whiteknight so if we take it out, it won't be able to build the attributes? 21:10
NotFound Is a data field, not a vtable function.
whiteknight: the attributes corresponding to the proxied parents. 21:11
cotto_work the build doesn't break if I take out the while loop. Util, this sounds like it's good a chance of being unneeded code.
whiteknight ok
Util Thanks, all. Will pursue. EOQ. 21:12
cotto_work we'll want to do a bit of digging before merging into master, but it looks promising
great. any other questions?
Did anyone have a question queued that I missed? 21:13
NotFound I had a question about nqp libraries, but was already answered in #parrot a few hours ago.
cotto_work ok. We have some goals this week: 21:15
GOAL 1: Get some GSoC ideas on the wiki. (all)
GOAL 2: Merge/review tewk/select. (cotto, tewk, others)
GOAL 3: evaluate attribute_defs, rip out in branch (util)
NotFound For the record: nqp libraries depend on P6object being loaded, and the answer is that it will be fixed. 21:16
whiteknight recorded
cotto_work any closing remarks? 21:17
tadzik I can say something about Da 21:18
... DaD, since there's no dukeleto 21:19
cotto_work tadzik: go ahead
dukeleto is here
tadzik so, I wrote a regex-based deprecation detector... oh, dukeleto's here
dukeleto tadzik: go ahead
tadzik: i am backlogging a bit
tadzik so, it works. Requires a bit extended api.yaml, and then matches given files for deprecations and outputs a friendly info 21:20
nqp script and api.yaml in github.com/tadzik/parrot-deprecation-detector
cotto_work tadzik: very nice
tadzik that's it
cotto_work tadzik: thanks for bringing attention to that 21:21
tadzik opens route to webservices and stuff
oh, Makefiles is broken
cotto_work tadzik: I'm looking forward to being able to tell users "run this script in your project's root dir and it'll detect most of the deprecations that you'll run into" 21:22
tadzik cotto_work: consider it done
say, next #ps
and that requires filling api.yaml with deprecations
I mean, with regexes 21:23
NotFound What kind of regexes?
tadzik hold on...
as in github.com/tadzik/parrot-deprecati...i.yaml#L18 21:24
dukeleto tadzik: p6 regexes ?
tadzik aye
it all runs on nqp
dukeleto tadzik: spiffy
tadzik spiffy?
dukeleto tadzik: spiffy = good stuff
tadzik oh, spiffy :) 21:25
cotto_work I like where all of this is going.
dukeleto i can imagine an automated process that checks out the latest versions of HLL's and runs the deprecation detector on them.
tadzik as spiffy as my Suceeded Driving Exam Today? :P
dukeleto: that's the goal methinks
dukeleto tadzik: congrats!
tadzik thanks :) 21:26
NotFound tadzik: So you are not Sponge Bob?
tadzik *SCNR*
21:26 whiteknight left
tadzik NotFound: no, I don't really like jeans 21:26
dukeleto tadzik: does the dep detector have docs?
tadzik dukeleto: nope
I can write some
dukeleto tadzik: we need some :)
tadzik: just very quick intro, explaining what it is and what it can do
tadzik currently, there's not much code
oh, a README or something. Sure thing
dukeleto tadzik: explain that it uses nqp-rx regexen, which a few simple examples
tadzik will do it probably tomorrow 21:27
dukeleto tadzik: yes, a README is very nice
tadzik:
tadzik a wiki page or something
dukeleto tadzik++
tadzik lemee add it to my todo
NotFound Yes, be very clear that they aren't perl5 ones.
Premature confusion avoiding is the root of all good.
cotto_work I'd think that in the common case, a full regex won't be necessary and that a simple string would be enough. 21:28
tadzik sanesounding
will think about it
cotto_work the lazy developer in me says that people probably won't want to figure out a p6regex if they don't absolutely have to 21:29
21:30 plobsing left
tadzik true 21:30
cotto_work This discussion should move to #parrot.
tadzik well, that's a 3.3 Goal 21:31
but maybe
cotto_work Are there any other #ps issues or should we call it a wrap? 21:34
That's a wrap! Go write great things. 21:36
NotFound say "great things"
kid51 cosas magnificas 21:37
tadzik Girrafes have 18-inch tongues, and they sleep 15mins-2hours daily
moderator Priorities for this week: irclog.perlgeek.de/parrotsketch/201...#i_3348283 | 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/ 21:38
21:39 NotFound left 21:42 atrodo left 21:53 plobsing joined 21:54 plobsing left 21:56 kid51 left 23:20 lucian_ joined 23:23 lucian left 23:39 whiteknight joined 23:52 lucian_ left 23:53 lucian 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!