Camelia, the Perl 6 bug

IRC log for #parrot, 2010-11-17

Parrot | source cross referenced

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:01 slavorgn left #parrot
00:05 cotto_work fortunately the built works fine from the tarball, though I'm sure the installed parrot will exhibit some quirks
00:08 slavorgn joined #parrot
00:12 bacek_at_work aloha, humans
00:13 cotto_work hi bacek_at_work
00:14 bacek_at_work hi cotto_work
00:15 bacek_at_work Does anyone know good perl module to round up time delta in human readable format? E.g. 70s -> '1 minute ago', 4000s -> 'About 1 hour ago' and so on.
00:20 Tene bacek_at_work: TimeDate::Format::Natural::Duration might be a good place to start.
00:20 Tene iirc
00:21 bacek_at_work Tene, thanks!
00:21 dukeleto we need a 2.10.1 and we need a volunteer to do it. Who will get the massive karma?
00:22 bacek_at_work Tene, looks like it's other way around.
00:23 cotto_work What should the git-related config keys be in a release?
00:24 Tene bacek_at_work: http://search.cpan.org/~dmuey/DateT​ime-Format-Human-Duration-0.0.1/lib​/DateTime/Format/Human/Duration.pm
00:25 bacek_at_work Tene, yeah. Thanks, I'll look at it.
00:45 dalek parrot: b77b0f2 | plobsing++ | tools/dev/nci_thunk_gen.nqp:
00:45 dalek parrot: dead code removal
00:45 dalek parrot:
00:45 dalek parrot: This was added with gsoc_nci. However, that portion of gsoc_nci was not
00:45 dalek parrot: completed and has been culled. This is not the appropriate tool for creating
00:45 dalek parrot: nci thunks at the current time.
00:45 dalek parrot: review: https://github.com/parrot/parrot/commit/​b77b0f2d37d0d6e15714c8cdb734bf35b998b5fd
00:45 dalek parrot: 980409b | plobsing++ | MANIFEST:
00:45 dalek parrot: mk_manifest_and_skip
00:45 dalek parrot: review: https://github.com/parrot/parrot/commit/​980409b86f128b283346d85959eb3abcce6e0acc
00:45 dalek parrot: f6ca948 | plobsing++ | NEWS:
00:45 dalek parrot: add gsoc_nci merge to news
00:45 dalek parrot: review: https://github.com/parrot/parrot/commit/​f6ca9482c0a81fb982f685caf7dfcde20cbd9a2e
01:36 bacek_at_work msg Coke You should have some info about message age from aloha
01:36 aloha OK. I'll deliver the message.
01:36 bacek_at_work msg bacek_at_work blah
01:36 aloha OK. I'll deliver the message.
01:36 bacek_at_work ~~
01:46 cotto ~~
01:50 seanstickle left #parrot
02:13 mikehh dukeleto: the message in the tarball is spurious - it goes on to build fine.
02:16 mikehh cotto: the error message generated in configure is spurious - it goes on to build with no problems and PASSes all tests (up to fulltest)
02:34 rurban_ joined #parrot
02:38 rurban left #parrot
02:38 rurban_ is now known as rurban
02:48 plobsing ping rurban
03:01 bluescreen left #parrot
03:04 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#1135) fulltest) at f6ca948 - Ubuntu 10.10 i386 (g++-4.5 with --optimize)
03:33 jsut joined #parrot
03:34 dngor left #parrot
03:37 jsut_ left #parrot
03:45 dngor joined #parrot
03:47 aloha left #parrot
03:48 aloha joined #parrot
03:52 preflex left #parrot
03:56 preflex joined #parrot
03:57 dngor left #parrot
04:09 dngor joined #parrot
04:37 seanstickle joined #parrot
04:53 lidden joined #parrot
05:03 dukeleto mikehh: hmmm, perhaps we don't need a new tarball then.
05:04 seanstickle left #parrot
06:18 ZeroForce joined #parrot
06:20 ZeroForce left #parrot
06:27 bacek_at_work left #parrot
06:27 bacek_at_work joined #parrot
06:32 theory left #parrot
07:18 fperrad joined #parrot
07:21 bacek joined #parrot
08:01 dalek parrot: d662d44 | plobsing++ | DEPRECATED.pod:
08:01 dalek parrot: deprecate pkg-config support
08:01 dalek parrot: review: https://github.com/parrot/parrot/commit/​d662d4448c3d035607f7ab2c6eb8cf89aa2c08ab
08:13 dalek TT #1853 created by plobsing++: pkg-config support
08:13 dalek TT #1853: http://trac.parrot.org/parrot/ticket/1853
08:23 bacek left #parrot
08:29 dalek tracwiki: v2 | plobsing++ | ParrotDeprecationsFor3.0
08:29 dalek tracwiki: http://trac.parrot.org/parrot/wiki/ParrotDe​precationsFor3.0?version=2&action=diff
08:29 dalek tracwiki: v18 | plobsing++ | ParrotDeprecations
08:29 dalek tracwiki: http://trac.parrot.org/parrot/wiki/Parro​tDeprecations?version=18&action=diff
08:46 bacek joined #parrot
08:55 moritz GeJ: has https://github.com/perl6/nqp-rx/pull/10 been taken care of? the link to the commit is a 404
08:56 moritz hm, has been closed
08:58 silug left #parrot
09:25 jjore left #parrot
09:27 he joined #parrot
09:28 bacek left #parrot
09:38 jjore joined #parrot
10:03 bacek joined #parrot
10:28 dalek parrot/gc_ms2_faster_list: 111d25f | bacek++ | src/gc/gc_ms2.c:
10:28 dalek parrot/gc_ms2_faster_list: Remove usused static functions
10:28 dalek parrot/gc_ms2_faster_list: review: https://github.com/parrot/parrot/commit/​111d25f07f11f806615db9f2301db1f856207f53
10:28 dalek parrot/gc_ms2_faster_list: 525f058 | bacek++ | src/gc/gc_ms2.c:
10:28 dalek parrot/gc_ms2_faster_list: Fix compiler warning
10:28 dalek parrot/gc_ms2_faster_list: review: https://github.com/parrot/parrot/commit/​525f058e8726769ef099a132cd968c88f5875c70
10:28 dalek parrot/gc_ms2_faster_list: 36be44a | bacek++ | src/gc/gc_ms2.c:
10:28 dalek parrot/gc_ms2_faster_list: Fix putting freed PMC back to Pointer_Array
10:28 dalek parrot/gc_ms2_faster_list: review: https://github.com/parrot/parrot/commit/​36be44a2d6c6a99666fc55c96cea0dfc7149abc3
10:28 dalek parrot/gc_ms2_faster_list: 17d7fe2 | bacek++ | src/gc/gc_ms2.c:
10:28 dalek parrot/gc_ms2_faster_list: Fix and reenable sweeping of string pool.
10:28 dalek parrot/gc_ms2_faster_list: review: https://github.com/parrot/parrot/commit/​17d7fe2d8c089d78899715ed1af296e6c1ae68ea
10:35 rurban_ joined #parrot
10:38 rurban left #parrot
10:38 rurban_ is now known as rurban
10:39 bacek aloha, ( (11*60 + 16) - (9 * 60 + 40) ) / (11*60 + 16) * 100
10:39 aloha bacek: 14.2011834319527
10:41 sorear sexagesimal--
10:52 dalek parrot/gc_ms2_faster_list: d93e81a | bacek++ | src/gc/gc_ms2.c:
10:52 dalek parrot/gc_ms2_faster_list: Remove more unused functions.
10:52 dalek parrot/gc_ms2_faster_list: review: https://github.com/parrot/parrot/commit/​d93e81a8415f0f7f02f54ae2ca26a0fb7d31cc31
10:52 dalek parrot/gc_ms2_faster_list: ad13215 | bacek++ | src/gc/gc_ms2.c:
10:52 dalek parrot/gc_ms2_faster_list: Reimplement destroy_pmc_pool function
10:52 dalek parrot/gc_ms2_faster_list: review: https://github.com/parrot/parrot/commit/​ad1321533ca16e8f97764602355c6abf479dd2c8
10:52 dalek parrot/gc_ms2_faster_list: fc3e87b | bacek++ | src/gc/gc_ms2.c:
10:52 dalek parrot/gc_ms2_faster_list: Fix codetest
10:52 dalek parrot/gc_ms2_faster_list: review: https://github.com/parrot/parrot/commit/​fc3e87b7a173cac5ad7dcffcfd7600a2d8675a09
11:31 bacek aloha, (501 - 276) / 501 * 100
11:31 aloha bacek: 44.9101796407186
12:09 lidden left #parrot
13:18 confound left #parrot
13:18 confound joined #parrot
13:20 PerlPilot joined #parrot
13:21 seanstickle joined #parrot
13:25 PerlJam left #parrot
13:25 PerlJam joined #parrot
13:27 PerlPilot left #parrot
13:30 whiteknight joined #parrot
13:42 bluescreen joined #parrot
13:42 whiteknight seen tewk?
13:42 aloha Sorry, I haven't seen tewk.
13:48 whiteknight msg chromatic: Is there anything worth saving in the tt389_fix branch? I don't think any of the work I was doing there is worth keeping (good idea, but wrong implementation direction). We can delete it if you don't need anything from it
13:48 aloha OK. I'll deliver the message.
13:52 whiteknight msg chromatic all my testing in the gc_ms2_sf branch is coming back clean. Do you want to merge it?
13:52 aloha OK. I'll deliver the message.
14:10 whiteknight good morning, #parrot
14:13 lucian joined #parrot
14:21 dalek parrot: dbc1e22 | fperrad++ | tools/dev/fetch_languages.pl:
14:21 dalek parrot: [languages] many languages move on github/parrot
14:21 dalek parrot: review: https://github.com/parrot/parrot/commit/​dbc1e22038568c0674fd560d24e3a56af1df50d5
14:31 gerd joined #parrot
14:34 gerd There is a problem to run "perl Configure.pl" by release 2.10.0
14:34 gerd The output is:
14:34 dalek parrot: 3117731 | fperrad++ | tools/dev/fetch_languages.pl:
14:34 dalek parrot: [languages] missing ook
14:34 dalek parrot: review: https://github.com/parrot/parrot/commit/​3117731d6d70a5f6f55ca1e2bf4e7f6c45bb856b
14:35 gerd auto::ctags -         Is (exuberant) ctags installed......................yes.
14:35 gerd auto::revision -      Determine Parrot's revision...........................1.fatal: Not a git repository (or any of the parent directories): .git
14:35 gerd auto::sha1 -          Determine Parrot's sha1...
14:35 gerd step auto::sha1 died during execution: Invalid Parrot sha1 (SHA1):  at config/auto/sha1.pm line 37.
14:35 gerd at Configure.pl line 74
14:35 gerd fatal: Not a git repository (or any of the parent directories): .git
14:35 gerd auto::git_describe -  Determine Parrot's git-describe...
14:36 gerd It think it would be best to set the revision to 0 for releases
14:37 whiteknight that's still an issue? I thought somebody was going to fix that yesterday
14:38 gerd I just downloaded the tar file and tried to build
14:41 whiteknight gerd: I'm reading the backlog now. mikehh says that it should still build fine, even with the error messages
14:41 whiteknight can you try make?
14:43 gerd make runs
14:43 whiteknight okay
14:43 whiteknight bacek: ping
14:45 fperrad msg dukeleto github.com/parrot/ecmascript is not synchro with svn.parrot.org/languages/ecmascript
14:45 aloha OK. I'll deliver the message.
14:45 bluescreen left #parrot
14:50 gerd the error from "perl Configure.pl" stops building RPMs
14:52 gerd A workaround would be possible but a new tar file (2.10.1) would be nice
14:53 whiteknight yes, a new tar file would be nice
14:54 he Hmm, I'm trying to build from an extracted tar file, and configure fails with "auto::revision -      Determine Parrot's revision...........................1.fatal: Not a git repository (or any of the parent directories): .git"
14:54 he (2.10.0)
14:55 he Ah, I see others are seeing the same.
14:59 gerd whiteknight: should I start to make 2.10.1 from the current git repository, I do not know how fast I can do this? I two hours I will have to fetch my doughter
14:59 he ...and .parrot_current_sha1 is near-empty (single newline only), and configure dislikes that as well.
14:59 PerlJam So ... what's the fix for lack of a git repo?  the tarball should still know what revision it was made from, yes?
15:00 he I would have thought so.  Embed the information when the release is cut?
15:00 bluescreen joined #parrot
15:00 * moritz would be fine with having the VERSION_2_10_1 string in the git_describe config field
15:01 moritz erm, s/VERSION/RELEASE/
15:02 moritz gerd: I think it would make more sense to start a branch off the previous release, and cherry-pick dukeleto++'s configure fixes
15:02 moritz and release the result
15:03 gerd Okay I will try to this.
15:07 gerd if it is not done in the next two  hours if I will finish it at home after fetching my doughter
15:09 Coke as noted in the ticket, the tarball will work or not work depending on whether or not you have git installed.
15:09 Coke bacek++
15:19 dalek parrot: ae5b702 | plobsing++ | src/dynoplibs/deprecated.ops:
15:19 dalek parrot: remove exchange dynop
15:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​ae5b70220b77b71411230ddfa1c43f356fc95717
15:19 dalek parrot: 7151902 | plobsing++ | t/dynoplibs/deprecated.t:
15:19 dalek parrot: remove exchange op tests
15:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​71519028cd098d327eb7c8d5d65edd090c1c04d0
15:22 dalek TT #1643 closed by plobsing++: [DEPRECATION] exchange op
15:22 dalek TT #1643: http://trac.parrot.org/parrot/ticket/1643
15:36 JimmyZ joined #parrot
15:37 mikehh left #parrot
15:38 darbelo joined #parrot
15:42 seanstickle left #parrot
15:44 JimmyZ left #parrot
15:45 theory joined #parrot
15:46 stilgar joined #parrot
15:49 theory left #parrot
15:53 dmalcolm joined #parrot
16:08 dukeleto 'ello
16:10 cotto 'i
16:11 gerd Hello
16:13 dukeleto So are we still planning on making a 2.10.1 ?
16:13 * dukeleto hasn't backlogged
16:14 moritz dukeleto: gerd is
16:14 Coke +1, it needs doing.
16:15 gerd Also in the current revision is the Configure problem
16:15 moritz dukeleto: is 4820ad51b936d2a90639e8b4ef512a66b40bdc1f the only commit necessary in 2.10.1?
16:18 dukeleto moritz: that fixes one bug, but doesn't fix the issue that stuff will fail if git is not available
16:18 moritz dukeleto: that's not fixed in master at all?
16:20 dukeleto moritz: i haven't backlogged and I have been doing other things since I was last here, so unless someone else did it, no.
16:20 moritz dukeleto: ok, thanks
16:22 dukeleto moritz: it shouldn't take long to fix, but i just didn't have time yesterday
16:23 moritz dukeleto: I didn't mean to make any pressure, just wanted to know the current state of affairs
16:27 gerd should I make a 2.10.1 with a workaround that hardcode "RELEASE_2_10_1"
16:28 dukeleto gerd: huh?
16:28 dukeleto moritz: i think i know how to fix it. I will make the config steps do nothing if the "DEVELOPING" file is not present. Does that sound sane to you?
16:29 dukeleto moritz: should sha1 and git_describe just be empty keys, in that case?
16:29 moritz sha1 should be empty
16:30 moritz git_describe could hold the tag name
16:30 moritz or be empty
16:30 moritz both work for me
16:31 dukeleto moritz: both will be empty in the case of "git" not being installed, and we don't want git to need to be installed for tarballs to work
16:31 moritz ok.
16:31 dukeleto moritz: where would the tag name come from if git is not installed?
16:31 dukeleto moritz: if we cache it somewhere, then that is fine. But we can't ask "git" anything from a tarball
16:31 Coke I think andy's suggestion on the ticket is a sane one.
16:32 dukeleto Coke: link?
16:32 Coke that it key off .git instead of DEVELOPING, as we might have nightly tarballs at some point, etc.
16:32 moritz +1 to that too
16:32 dukeleto Coke: yes, that is smart.
16:33 moritz dukeleto: well, if version eq git tag, we have that information available. But it's really no big deal for me if git_describe is empty
16:35 gerd I think the version has to come from the VERSION file on the tarball is no git
16:36 dukeleto i dont' see any special handling of this case in Parrot::Revision. Wouldn't this same bug happen if people didn't have "svn" installed when using a tarball?
16:36 Coke never came up.
16:36 dukeleto Coke: you have a link for the ticket that andy mentions using .git ?
16:36 Coke so I'm guessing not.
16:37 Coke not without digging through commit mail, no.
16:37 Coke s/commit/ticket/
16:37 Coke https://trac.parrot.org/pa​rrot/ticket/1852#comment:7
16:37 gerd by svn the revison was set to 0 on release tarballs
16:40 dukeleto was the DEVELOPING file in the 2.10.0 release tarball?
16:41 gerd no, it is not
16:42 dukeleto good.
16:42 * dukeleto fixes the git bugs
16:43 dukeleto incoming!
16:43 dalek parrot: a5ba11a | dukeleto++ | config/auto/ (2 files):
16:43 dalek parrot: [config] Make git-related config steps do nothing unless we are in a Git repo
16:43 dalek parrot: review: https://github.com/parrot/parrot/commit/​a5ba11ab5c0bc6f42706ac38ed442b269ebb3e56
16:44 dukeleto I manually tested it, and it seemed to work.
16:44 gerd I have to leave for fechting my doughter, after that I will look here again, good luck
16:44 gerd left #parrot
16:49 jan joined #parrot
17:01 theory joined #parrot
17:04 dukeleto moritz: that commit above should fix the "git from a tarball" issue
17:05 dukeleto moritz: so the answer to your previous question is "both those commits need to be in 2.10.1"
17:05 moritz ok
17:05 moritz need to decommute now
17:05 moritz dukeleto++
17:21 whiteknight gc_ms2_faster_list is pretty cool, and a nice surprise
17:21 whiteknight 15% on rakudo build is nothing to sneeze at
17:21 whiteknight I wonder how it's going to interact with gc_ms2_sf, which boasts ~10% speedups in some tests as well
17:25 atrodo I should create http://isparrotfastyet.com/
17:25 dukeleto Anybody ever heard of the "Clay" language? It seems to be mirrored on github now: https://github.com/jckarter/clay and is built on top of LLVM
17:26 whiteknight dukeleto: never heard of it.
17:27 whiteknight atrodo: effort better spent improving speed in Parrot!
17:28 atrodo whiteknight> you're probably right
17:28 hercynium left #parrot
17:29 cotto_work atrodo, if you want to buy that domain and put up a clone of speed.pypy.org you could do both
17:30 whiteknight if these two GC branches come in and give us ~20% improvement, that will be a nice big jump
17:30 whiteknight Not to mention the generational_gc work, which should do just as much and be additive
17:30 dukeleto atrodo: yes, i have many ideas for a benchmarking site for parrot.
17:30 atrodo dukeleto> i'm curious to hear them
17:31 atrodo cotto_work> I would model it off of http://arewefastyet.com/
17:31 atrodo cotto_work> but the pypy site is cool as well
17:32 dukeleto "Clay has the same memory footprint and runtime overhead as C. It is suitable for writing garbage collectors, embedded systems, database servers, games etc. "
17:32 dukeleto That is an interesting niche.
17:34 cotto_work atrodo: you mean the one that says "502 Bad Gateway" on its front page?
17:36 whiteknight cotto_work:  it's a metaphor. If you don't need to get the correct answer, you can make it extremely fast
17:39 dukeleto whiteknight: have you read http://wiki.pentaho.com/display/​BEEKEEPER/3.+The+Beekeeper+Model ? It is mentioned many good principles of product management
17:40 dukeleto s/It is mentioned/It mentions/
17:41 whiteknight dukeleto: no, I have never read that
17:41 dukeleto whiteknight: it is a good read
17:45 whiteknight yes. I'm skimming it now
17:53 moritz gerd just told me that real life got in his way to make the 2.10.1 release
17:53 moritz any other volunteers?
17:53 moritz I'm afraid I'll be too distracted to do it myself
17:55 plobsing_ joined #parrot
17:56 AzureStone left #parrot
17:57 AzureStone joined #parrot
17:58 dalek parrot/prepare_2_10_0: 7240365 | dukeleto++ | config/auto/ (2 files):
17:58 dalek parrot/prepare_2_10_0: [config] Make git-related config steps do nothing unless we are in a Git repo
17:58 dalek parrot/prepare_2_10_0: review: https://github.com/parrot/parrot/commit/​7240365cbf1ac9709f17f72b949d09b83b468027
17:59 moritz I've pushed the prepare_2_10_0 branch, which should have all the necessary fixes
18:02 dukeleto moritz: so what do you need?
18:03 dukeleto moritz: that branch is basically what we need to tag as 2.10.1 ?
18:03 plobsing_ question regarding deprecation process: according the steps currently on the wiki, entries in DEPRECATED.pod don't get removed or updated after a deprecated item is complete. won't it become difficult over time to determine which items are still active?
18:03 dukeleto plobsing_: that is a bug in the docs. Stuff gets removed from DEPRECATED.pod when it is removed
18:03 dukeleto plobsing_: the info moves to the appropriate deprecation wiki page
18:04 plobsing_ OK, updating docs.
18:05 dukeleto plobsing_++
18:10 dalek parrot: 949f1b9 | plobsing++ | DEPRECATED.pod:
18:10 dalek parrot: remove deprecation notice for exchange op
18:10 dalek parrot:
18:10 dalek parrot: this has been completed
18:10 dalek parrot: review: https://github.com/parrot/parrot/commit/​949f1b97fc1195c1ec6398269f9879e426f59f36
18:10 joekonno joined #parrot
18:10 cotto_work plobsing++ for decrufitication
18:11 joekonno is now known as tojek
18:13 tojek left #parrot
18:15 dalek tracwiki: v7 | plobsing++ | HowToDeprecate
18:15 dalek tracwiki: prevent DEPRECATED.pod from filling up with old entries
18:15 dalek tracwiki: http://trac.parrot.org/parrot/wiki/How​ToDeprecate?version=7&action=diff
18:15 tojek joined #parrot
18:16 dukeleto so, who wants to be release manager for 2.10.1 ? I think moritz did all the git-related work, if there is any left, I can do it. We need someone for the rest of the process.
18:19 cotto_work I'll take it if it's not done by the time I get home tonight.
18:33 cotto_work not parrot related but still pretty cool: http://singularityhub.com/2010/11/17​/ai-sports-journalist-covers-every-d​ivision-i-college-basketball-team/
18:35 rurban_ joined #parrot
18:38 tojek left #parrot
18:38 hercynium joined #parrot
18:38 rurban left #parrot
18:38 rurban_ is now known as rurban
18:54 mikehh joined #parrot
18:58 plobsing_ left #parrot
19:07 atrodo cotto_work> strange, the link worked for me
19:11 Coke what is the point of auto::revision now?
19:12 dukeleto Coke: back-compat
19:13 dukeleto Coke: every HLL out there asks parrot_config for it's revision when building, to see if a new enough parrot is available
19:13 dukeleto Coke: so we need to keep the revision key around for a bit, so that stuff fails properly, and when we have the new tools ready to build HLL's on parrot.git, we will deprecate and remove it
19:14 dukeleto Coke: we probably need to add "parrot_config revision" as a deprecation that will be removed
19:14 dukeleto Coke: "fails properly" == "HLL's ask for a revision and think they are too old"
19:14 dukeleto instead of blowing up because a revision key does not exist
19:17 cotto_work dukeleto: once they're sufficiently culled, perhaps we should start covering parrot_config keys in the support policy.
19:18 Coke dukeleto: given that I'm failing if it's too old, it works out to the same thing. but ok.
19:19 cotto_work currently it smacks of a type of thinking that says "that value is easy to add; we should add that"
19:23 dukeleto Coke: not really. An HLL script that knows it is too old and stops to inform the user is different than an HLL that blows up at configure time.
19:23 dukeleto Coke: the perception from the end user is much different.
19:27 bluescreen left #parrot
19:28 Coke dukeleto: ... In both cases, I will stop and say "your parrot is too old."
19:29 Coke oh, I see what you're saying.
19:32 Coke Might be worth putting a note in auto:revision about that.
19:37 bluescreen joined #parrot
19:38 cotto_work hi bluescreen
19:38 bluescreen hi
19:40 dalek parrot: 410b2bf | Coke++ | config/auto/git_describe.pm:
19:40 dalek parrot: Shorten the description, make output fit on 1 line.
19:40 dalek parrot: (The value returned by this config step is very long.)
19:40 dalek parrot: review: https://github.com/parrot/parrot/commit/​410b2bf46ec254f82a756c59b0a7d2e5e06bdc05
19:42 bluescreen this is too funny http://digitizor.com/2010/11/17/internet-explo​rer-9-caught-cheating-in-sunspider-benchmark/ ... I would seek for another job rather then write code for have a high mark in an specific benchmark
19:43 bacek Good morning
19:43 bacek whiteknight, pong
19:44 whiteknight bacek: have you tested the new gc_ms2_faster_list branch with the changes made in gc_ms2_sf?
19:44 bacek whiteknight, nope.
19:44 whiteknight okay, that branch is going to be merged soon too
19:45 bacek Good. We should have substantial performance improvements after merging both branches.
19:46 whiteknight okay. I think _sf could merge today. I haven't talked to chromatic, but I think I could merge that branch tonight
19:46 whiteknight it has passed all my tests
19:48 cotto_work bluescreen: MS culture FTW
19:50 cotto_work Hopefully the exposure will motivate them in the right direction.
19:51 whiteknight bluescreen: I'm 95% done getting the embedding API to build. I have a few problems left with flags
19:51 whiteknight When it works, I'm going to push a copy to a branch in parrot/parrot.git
19:57 cotto_work at least you know they care about their benchmark numbers
19:58 atrodo What is the best methodology to benchmark parrot?  time make test?
20:00 TimToady phone?
20:00 dukeleto atrodo: examples/benchmarks
20:00 dukeleto atrodo: oofib.pir is a good benchmark for general performance
20:00 dukeleto atrodo: there are other benchmarks to test specific subsystems
20:01 atrodo dukeleto> oh, very nice.  didn't know those were there
20:01 whiteknight we can always use more benchmarks too, if you're feeling industrious
20:02 cotto_work atrodo: the nqp-rx build is good too
20:02 atrodo but what about calculating timings?
20:05 whiteknight the benchmarks themselves should display timing info, I think
20:09 dukeleto atrodo: i am working on this https://github.com/notbenh/tool_bench with notbenh
20:09 dukeleto atrodo: we need docs, but basically, it is a general benchmark framework. you give it lists of interps and files and it does all the work and give you back your data in the format of your choice
20:09 dukeleto atrodo: currently JSON and text formats are supported
20:10 dukeleto atrodo: it evolved from this https://github.com/notbenh/euler_bench
20:10 atrodo dukeleto> nice.  I'm going to have to play
20:10 dukeleto atrodo: please do. We need some end-users of tool_bench to make us write docs :)
20:11 dukeleto atrodo: notbenh is usually in #pdx.pm, you can ask him questions there :)
20:11 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#1153) fulltest) at 410b2bf - Ubuntu 10.10 amd64 (g++-4.5)
20:13 whiteknight yay! A pull request from a new user
20:13 atrodo dukeleto> Unfortunately, I may be a bad end user since I'm reading the code as documentation
20:13 dukeleto atrodo: no, you are the best end user
20:14 dukeleto atrodo: the perl modules have examples in POD, that should get you started
20:14 atrodo dukeleto> Thanks, I completely ignored the pod example
20:14 dukeleto atrodo: please let me know if it works for you. I have been slowly working on that, and will use it in my mythical parrot benchmarking website
20:14 * whiteknight really wishes our "ecmascript" project had a nicer, sexier name
20:15 whiteknight "ecmascript" is so damn sterile
20:16 atrodo dukeleto> well, i'm contemplating registering http://isparrotfastyet.com/ and setting up a small benchmark-over-time site
20:16 atrodo whiteknight> caffeinatedparrotscript?
20:16 dukeleto atrodo: i have written most of the code to do that, i just haven't glued it all together. If you want to work together, it will get done a lot faster.
20:17 atrodo dukeleto> what are you missing?
20:17 dukeleto atrodo: the time to glue stuff together
20:19 atrodo dukeleto> :D  You have your code so far on github or somewhere?
20:19 Coke gluesniffer.
20:21 notbenh joined #parrot
20:21 dukeleto atrodo: tool_bench is a critical piece. and I have other critical pieces scattered across other github repos
20:22 notbenh hi all
20:22 Coke ~~
20:22 notbenh I see that the topic of the hour is tool_bench
20:22 notbenh any thing that I can add to the conversation?
20:23 dukeleto notbenh: atrodo and i want to build a parrot benchmarking site on top of tool_bench
20:24 notbenh dukeleto: gravy!
20:24 dukeleto atrodo: the idea that notbenh and I have been nurturing is this: we can write abstract software for implementing a "benchmark website". The detail of what you are benchmarking is irrelavant
20:24 notbenh are you planning on just having some post-commit hook that runs a rev?
20:24 dukeleto I think having a passive pull occasionally is the way to go.
20:24 notbenh Currently tool_bench (and thus bench) have JSON output
20:24 dukeleto but we can do a post-commit hook as well.
20:25 notbenh dukeleto: well how it's tripped is some what irrelivant
20:25 notbenh ya if you just need the timing data then what we have now should at least give you that
20:25 notbenh though if you can think of specific features I'll gladly add them
20:26 notbenh well add them some where
20:26 dukeleto notbenh: i think tool_bench mostly needs docs and examples now
20:26 notbenh dukeleto: ya
20:26 dukeleto notbenh: the next layer on top of tool_bench needs to be implemented
20:26 notbenh and some outside tool_bench config => time thing
20:26 dukeleto notbenh: which read a config file and then farms that out to tool_bench
20:26 notbenh ya
20:27 notbenh as long as we aggree that 1) yes it needs to be built and 2) it's not tool_benches job to read that config then were both on the same page
20:27 notbenh dukeleto: did you want to just start with the eb config? or were there shortcomings that should be addressed now
20:27 dukeleto notbenh: yep. tool_bench is feature-complete, except for more output plugins
20:28 dukeleto notbenh: the euler_bench config file is probably a good starting point
20:28 notbenh dukeleto: just make a new package Tool::Bench::Report::MyFormat and then pass report 'MyFormat' and it'll find it
20:29 theory left #parrot
20:30 dukeleto notbenh: i am happy with JSON, just saying that is where new features will go
20:30 notbenh dukeleto: ahh
20:34 bluescreen left #parrot
20:35 bluescreen joined #parrot
20:38 bacek whiteknight, did you test rakudo's build time on gc_ms2_sf?
20:41 whiteknight bacek: I have not. chromatic said he was seeing 3%-10% for most tasks
20:43 bacek whiteknight, I've got very bad feelings...
20:43 whiteknight why so?
20:44 bluescreen whiteknight++ 95% done, replacing also parrot.exe ?
20:45 bluescreen I wasn't sure what to do with the globals when you have a parent interpreter, should we copy globals to children interpreters... ?
20:45 whiteknight bluescreen: that's a much bigger question than just the embedding API
20:45 bacek whiteknight, i tried to merge _sf & _faster_list. And it's slower overall...
20:46 whiteknight bacek: okay. That's weird. we need to figure out why that is
20:46 whiteknight which of the two brings the biggest gains?
20:46 whiteknight we pick whichever once is best, merge that to trunk, then work on the other one separately
20:46 bacek _faster_list
20:47 whiteknight okay, focus on that. When can we merge it?
20:47 bacek after testing
20:47 bacek it's ready
20:49 Coke seen allison?
20:49 aloha allison was last seen in #parrot 23 hours 55 mins ago saying "atrodo: confusingly, the plural is spelled the same".
20:49 whiteknight bacek: building now. will try to test asap
20:49 Coke aloha, the plural of atrodo is atrodo.
20:49 aloha Coke: Okay.
20:49 bluescreen whiteknight, we should deprecate the old API... tough I'm not sure how the procedure is for doing such thing
20:49 bacek whiteknight, ok
20:49 whiteknight bluescreen: I don't want to deprecate it until we have a replacement available for review
20:50 bluescreen and even so, they should live together for a while allowing users to migrate from one to the other
20:50 whiteknight yes, exactly
20:50 Coke Suggestion for deprecations: in addition to adding it to the DEPRECATIONS.pod, send out an email to parrot-users with a link to the ticket.
20:50 whiteknight all the new api functions are named "Parrot_api_*", which is new and doesn't conflict with anything else
20:50 bluescreen is there any TT#?
20:50 whiteknight bluescreen: not yet. You want to create one?
20:51 bluescreen will do
20:51 allison Coke: :)
20:52 bacek aloha, nopaste?
20:52 aloha bacek: nopaste is is http://nopaste.snit.ch (works with the script in $_PARROT/tools/dev/nopaste.pl)
20:54 whiteknight bluescreen: assign it to me please
20:55 bluescreen sure
20:55 * whiteknight is stupid
20:56 whiteknight I've spent two hours staring at this data persistance bug. I'm calling all the right methods, but nothing was saving to the DB
20:56 whiteknight never called .Commit()
21:01 bluescreen need Trac Admin give me permissions to create tickets
21:04 whiteknight what's your trac username?
21:06 whiteknight bluescreen: added
21:06 seanstickle joined #parrot
21:12 atrodo dukeleto> Sorry for disappearing, $dayjob and all
21:13 atrodo dukeleto> Do you have any website code written?
21:14 nopaste "bacek" at 192.168.1.3 pasted "Some benchmarks" (34 lines) at http://nopaste.snit.ch/25704
21:14 bacek whiteknight,  http://nopaste.snit.ch/25704
21:15 bluescreen left #parrot
21:16 darbelo left #parrot
21:16 darbelo joined #parrot
21:16 darbelo left #parrot
21:19 dukeleto atrodo: nope
21:20 * atrodo feels another rathole approaching...
21:21 whiteknight bacek: good enough for me
21:21 whiteknight bacek: my coretest just passed
21:21 whiteknight (win63)
21:21 whiteknight (win64)
21:21 whiteknight I will run more tests tonight when I get home
21:21 atrodo whiteknight> I think you were right the first time.  Windows is 1 bit short
21:29 bluescreen joined #parrot
21:29 whiteknight left #parrot
21:34 theory joined #parrot
21:40 Coke win63, win64: whatever it takes.
21:41 cotto_work dukeleto: have you looked at codespeed, which is what speed.pypy.org runs: https://github.com/tobami/codespeed
22:00 theory left #parrot
22:01 theory joined #parrot
22:04 bacek left #parrot
22:07 seanstickle left #parrot
22:15 dalek parrot: 37f8015 | gerd++ | lib/Parrot/Git/Describe.pm:
22:15 dalek parrot: make git_describe more sure
22:15 dalek parrot: review: https://github.com/parrot/parrot/commit/​37f801542db8ee98e8ba1cd311ad35c1e4f53439
22:18 dukeleto cotto_work: codespeed looks interesting, but is not for me
22:18 cotto_work dukeleto: why not?
22:19 dukeleto cotto_work: i want the data generation and website to be totally decoupled.
22:19 dukeleto theory: ping!
22:19 theory dukeleto: pong
22:20 dukeleto theory: i have questions about Test.More, where is a good place to ask those?
22:20 theory Erm. Not sure. /msg I guess. no channel for it.
22:20 theory Unless maybe there's a #tap channel or something? That'd make sense.
22:24 sorear go to #perl and track down Schwern and/or igny
22:24 sorear ingy
22:25 sorear (what's thw question?)
22:29 fperrad left #parrot
22:36 whiteknight joined #parrot
22:40 dukeleto sorear: i bump into schwern often enough, I try not to seek him out.
22:40 dukeleto sorear: and I am talking about the javascript lib Test.More that theory wrote, not Test::More
22:41 theory Yeah, but ingy has used it more than anyone else.
22:42 bluescreen left #parrot
22:48 donaldh joined #parrot
22:56 dmalcolm left #parrot
22:59 dalek tracwiki: v3 | plobsing++ | ParrotDeprecationsFor3.0
22:59 dalek tracwiki: add deprecation information about indirect register ops
22:59 dalek tracwiki: http://trac.parrot.org/parrot/wiki/ParrotDe​precationsFor3.0?version=3&action=diff
22:59 dalek tracwiki: v19 | plobsing++ | ParrotDeprecations
22:59 dalek tracwiki: add removal notice for deprecated indirect register access ops
22:59 dalek tracwiki: http://trac.parrot.org/parrot/wiki/Parro​tDeprecations?version=19&action=diff
23:05 bacek_at_work ~~
23:06 cotto_work ~
23:06 lucian left #parrot
23:07 dukeleto theory: good point.
23:07 bacek_at_work cotto_work, what are you doing at work at this time???
23:09 bacek_at_work plobsing, can you add examples of "indirect register opcodes" into wiki?
23:11 plobsing bacek_at_work: done
23:12 bacek_at_work plobsing, erm... I couldn't find it
23:13 cotto_work http://trac.parrot.org/parrot/wiki/ParrotDe​precationsFor3.0?action=diff&version=4
23:13 theory left #parrot
23:14 theory joined #parrot
23:15 cotto_work bacek_at_work: trying to untangle the convoluted control flow that drupal themes go through
23:15 dalek tracwiki: v4 | plobsing++ | ParrotDeprecationsFor3.0
23:15 dalek tracwiki: slight clarification about what is meant by 'indirect register access ops'
23:15 dalek tracwiki: http://trac.parrot.org/parrot/wiki/ParrotDe​precationsFor3.0?version=4&action=diff
23:17 cotto_work dukeleto: I'd like to set the PDS to 15:15 on the 5th.  Any objections?
23:18 dukeleto cotto_work: where is the link for what everyones times are?
23:18 dukeleto cotto_work: did i put myself down as busy?
23:19 cotto_work I think everyone's free, at least everyone who's also free at 1700.
23:19 dalek parrot: fc8d990 | plobsing++ | src/dynoplibs/deprecated.ops:
23:19 dalek parrot: remove deprecated indirect register ops
23:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​fc8d990e875722abd9a7eac3b5eb44aacfac031f
23:19 dalek parrot: 6644b54 | plobsing++ | t/dynoplibs/deprecated.t:
23:19 dalek parrot: delete tests for deleted indirect register ops
23:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​6644b5464cd5d49b5f7162746f7621d2ac220282
23:19 dalek parrot: 52c2e30 | plobsing++ | MANIFEST:
23:19 dalek parrot: mk_manifest_and_skip
23:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​52c2e30051e0c8a2cdd6ee7d8704aa6b89c9b259
23:19 dalek parrot: f9e38eb | plobsing++ | / (2 files):
23:19 dalek parrot: remove Parrot_clear_x()
23:19 dalek parrot:
23:19 dalek parrot: These were used to implement the now-removed clear_x ops.
23:19 dalek parrot: They should be removed for the same reasons the ops were removed.
23:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​f9e38eb27364e81bfdf86f9e6ac9f939f53d67d1
23:19 dalek parrot: 79397d3 | plobsing++ | src/dynoplibs/ (3 files):
23:19 dalek parrot: remove deprecated_ops dynops lib
23:19 dalek parrot:
23:19 dalek parrot: this is now empty and can be removed
23:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​79397d3c614fba737df3950b74b6b485c408ac65
23:19 dalek parrot: 5f3e845 | plobsing++ | MANIFEST:
23:19 dalek parrot: mk_manifest_and_skip
23:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​5f3e84565d2208d5cee7e5c611379cb4dc2e0488
23:19 dalek parrot: 0575b21 | plobsing++ | include/parrot/call.h:
23:19 dalek parrot: headerizer
23:19 dalek parrot: review: https://github.com/parrot/parrot/commit/​0575b2182e92efb6e3bc4a38328605f8d741e92a
23:20 dukeleto cotto_work: if i put myself down as free, it is fine with me
23:20 dukeleto cotto_work: this is dec 5th, right?
23:20 cotto_work yes
23:21 plobsing where is a good place to document "this is explicitly not supported, if you do this in a dynop, I *WILL* break your code"
23:21 plobsing ?
23:21 cotto_work that will unfortunately exclude bacek_at_work but nobody else
23:21 bacek_at_work aloha, developer summit?
23:21 aloha bacek_at_work: Search me, bub.
23:22 bacek_at_work stupid girl...
23:22 cotto_work http://www.doodle.com/q4nwdt9i5k929n2s
23:22 dukeleto bacek_at_work: what do we have to do to get you to a dev summit?
23:22 dukeleto bacek_at_work: it is *virtual*, it shouldn't be that hard.
23:22 bacek_at_work dukeleto, timezones are our enemy :)
23:23 bacek_at_work aloha, developer summit is http://www.doodle.com/q4nwdt9i5k929n2s
23:23 aloha bacek_at_work: Okay.
23:24 dukeleto bacek_at_work: if you could send and email or something to parrot-dev, the day before the developer summit, that we be awesome. We want to hear your concerns and ideas, even if you can't attend.
23:24 cotto_work +1
23:25 bacek_at_work dukeleto, hmm... I don't see any immediate ideas apart from "make parrot awesome" and "make gc do not suck". But I'm working on it. At least second part.
23:25 dalek parrot: b57aae5 | plobsing++ | DEPRECATED.pod:
23:25 dalek parrot: indirect register opcode deprecation completed
23:25 dalek parrot: review: https://github.com/parrot/parrot/commit/​b57aae520c096da11451ae36915868cd86a42bd1
23:31 dalek TT #1642 closed by plobsing++: [DEPRECATION] ops which modify registers that are not their direct ...
23:31 dalek TT #1642: http://trac.parrot.org/parrot/ticket/1642
23:37 hercynium left #parrot
23:39 dalek parrot: 4e15d15 | gerd++ | lib/Parrot/SHA1.pm:
23:39 dalek parrot: to make no noise in a release tarball
23:39 dalek parrot: review: https://github.com/parrot/parrot/commit/​4e15d1583e940e4732f4ab95057651d5a11d4685
23:46 Andy left #parrot
23:47 bacek_at_work whoo-hoo! http://nopaste.snit.ch/25720
23:49 cotto_work bacek_at_work wins a pizza
23:50 cotto_work bacek++
23:50 bacek_at_work hey! what about beer?
23:50 bacek_at_work a lot of beer!
23:50 cotto_work sure.  I'll send you an imaginary beer too.
23:50 cotto_work and a real one if I ever meet you
23:51 dukeleto Real beer is just a 90 degree (pi/2) rotation of imaginary beer.
23:52 GeJ what does the "(0avgtext+0avgdata 1333168maxresident)k
23:52 GeJ " represent?
23:52 cotto_work so imaginary beer and half a pi == real beer?
23:53 dukeleto cotto_work: s/and/multiplied by/
23:54 cotto_work Dang.  I don't know how to multiply drinks by dessert.
23:54 sorear GeJ: According to the documenation? 1.33 GB peak RSS
23:54 dukeleto cotto_work: you should get on that.
23:54 sorear but the numbers are fishy
23:55 GeJ sorear: that's a lot of RAM
23:55 dngor_ joined #parrot
23:56 * dukeleto just compiled node.js for the first time. This stuff is shiny.
23:57 dngor left #parrot
23:57 sorear GeJ: right. fishy.

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

Parrot | source cross referenced