Camelia, the Perl 6 bug

IRC log for #parrot, 2011-04-18

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:00 cosimo joined #parrot
00:44 rohit_nsit08 left #parrot
00:44 rohit_nsit08 joined #parrot
00:46 Coke left #parrot
00:50 Coke joined #parrot
00:54 marcel_r left #parrot
00:57 rohit_nsit08 left #parrot
01:11 cotto dukeleto, ping
01:11 bubaflub joined #parrot
01:14 bubaflub ~
01:29 whiteknight left #parrot
01:39 lucian left #parrot
01:53 kid51 left #parrot
01:59 bubaflub left #parrot
02:00 contingencyplan left #parrot
02:42 woosley joined #parrot
02:52 rohit_nsit08 joined #parrot
02:52 dalek parrot: c4d3f78 | petdance++ | src/pmc/imccompiler.pmc:
02:52 dalek parrot: Consting and annotating
02:52 dalek parrot: review: https://github.com/parrot/parrot/commit/c4d3f78df6
02:52 dalek parrot: 7190577 | petdance++ | / (5 files):
02:52 dalek parrot: Merge branch 'master' of github.com:parrot/parrot
02:52 dalek parrot: review: https://github.com/parrot/parrot/commit/7190577b7d
02:59 hudnix left #parrot
02:59 rohit_nsit08 left #parrot
03:00 rohit_nsit08 joined #parrot
03:06 Eduardow left #parrot
03:11 Eduardow joined #parrot
03:34 soh_cah_toa left #parrot
04:10 rohit_nsit08 left #parrot
04:13 dalek parrot: 46c2a2b | petdance++ | / (2 files):
04:13 dalek parrot: Fixed up some function annotations. Removed outdated #define. Use STRINGNULL for our null strings.
04:13 dalek parrot: review: https://github.com/parrot/parrot/commit/46c2a2b42a
04:51 petdance left #parrot
05:36 zby_home left #parrot
05:47 theory joined #parrot
06:12 giwi joined #parrot
06:12 giwi left #parrot
06:13 giwi joined #parrot
06:21 fperrad joined #parrot
06:25 rohit_nsit08 joined #parrot
07:20 Coke left #parrot
07:21 Coke joined #parrot
07:26 cosimo left #parrot
09:23 ShaneC1 left #parrot
09:28 JimmyZ joined #parrot
09:28 rohit_nsit08 left #parrot
09:29 rohit_nsit08 joined #parrot
09:36 rohit_nsit08 left #parrot
09:36 JimmyZ left #parrot
09:36 rohit_nsit08 joined #parrot
09:37 jrtayloriv left #parrot
09:56 woosley left #parrot
09:58 bacek left #parrot
10:50 rohit_nsit08 left #parrot
10:50 rohit_nsit08 joined #parrot
11:09 rohit_nsit08 left #parrot
11:14 ambs joined #parrot
11:26 bacek joined #parrot
11:40 hudnix joined #parrot
11:56 Psyche^ joined #parrot
11:56 Patterner left #parrot
11:56 Psyche^ is now known as Patterner
12:01 mikehh left #parrot
12:15 redicaps joined #parrot
12:17 kuku joined #parrot
12:17 kuku left #parrot
12:17 kuku joined #parrot
12:28 benabik left #parrot
12:51 whiteknight joined #parrot
12:51 bubaflub joined #parrot
12:51 whiteknight good morning, #parrot
12:52 bacek good evening, whiteknight
12:52 whiteknight 4l3x4nd3rDW
12:52 whiteknight blah, wrong window
12:52 moritz please change that password :-)
12:52 whiteknight it's a work one. changes in a month anyway
12:52 bacek looks like a password :)
12:53 moritz and now it's in the public logs :-)
12:53 whiteknight are we ready/able to merge packfile_wrap?
12:53 moritz but +1 to using such strong passwords :-)
12:53 bacek whiteknight, I have no objections on merge it :)
12:53 whiteknight I tested the hell out of it yesterday
12:54 whiteknight I didn't merge it last night because the rakudo spectests took too long
12:54 giwi left #parrot
12:54 bacek 1903.94user 75.16system 33:22.46elapsed 98%CPU (0avgtext+0avgdata 837552maxresident)k
12:54 bacek on my box
12:54 whiteknight what's normal for master?
12:54 bacek mmm...
12:55 bacek I think it's faster on master. Can't get exact values.
12:55 whiteknight I'm sure it's a little faster. I would be surprised if there was a huge difference
12:56 utsl_ is now known as utsl
12:57 bacek Let'me do a quick check on compiling core.pm
13:02 whiteknight for some value of "quick"
13:04 whiteknight bacek: do you want to do the merge, or should I?
13:05 bacek aloha, clock?
13:05 aloha bacek: LAX: Mon, 06:05 PDT / CHI: Mon, 08:05 CDT / NYC: Mon, 09:05 EDT / UTC: Mon, 13:05 UTC / LON: Mon, 14:05 BST / BER: Mon, 15:05 CEST / TOK: Mon, 22:05 JST / SYD: Mon, 23:05 EST
13:05 whiteknight which one are you, SYD?
13:05 bacek whiteknight, I think it's safer if you'll merge it :)
13:05 bacek yes, SYD
13:05 whiteknight okay, I'll start on it now
13:06 whiteknight I'll run fulltest before I push
13:06 whiteknight then Rakudo spectest after
13:07 whiteknight of course, I'm giving blood in 10 minutes, so I might not be in condition to do this intelligently either :)
13:11 bacek whiteknight, mmm... Kind of bad news. --gc=gms isn't faster then ms2 anymore. 161 vs 166 seconds.
13:11 whiteknight why, because of the extra cost of using write barriers with packfiles?
13:11 bacek dunno
13:11 whiteknight okay, well, we can't do anything about it now
13:11 bacek writebarriers by it self are cheap.
13:12 whiteknight yeah
13:12 whiteknight they are cheap to call, but may be expensive if we have to re-mark all the contents of a packfile
13:12 whiteknight we have to get smarter about packfiles
13:12 whiteknight and stop merging them all the time
13:12 bacek most likely because we are marking constants multiple times now
13:12 whiteknight yeah, that's my thought
13:13 whiteknight we'll fix it after the release
13:13 bacek it can be... hard.
13:13 whiteknight I'm sure it will be hard
13:14 whiteknight If we stop  merging and updating packfiles, we can stop marking constants so much
13:14 whiteknight and I think we can move to a smarter system after the release
13:15 bacek nope. If we have proper (fsvo) PackfileConstant PMC used internally
13:15 bacek than we can stop marking constants multiple times
13:15 whiteknight yes, that would be good too
13:16 whiteknight we are just out of time for fixing it now. We need 3.3 to be stable
13:16 bacek hmmm...
13:16 whiteknight and you need sleep!
13:16 whiteknight brb. giving blood
13:17 bacek other possibility - I didn't make realclean. And was using gms all the time
13:25 bacek msg whiteknight All clear. ms2 vs gms - 232 vs 166 actually.
13:25 aloha OK. I'll deliver the message.
13:43 kaneplusplus left #parrot
13:44 giwi joined #parrot
14:02 Eduardow left #parrot
14:08 Eduardow joined #parrot
14:12 kuku left #parrot
14:13 benabik joined #parrot
14:13 benabik ~~
14:17 Andy_ joined #parrot
14:18 Coke left #parrot
14:20 jaffa4 joined #parrot
14:21 davidfetter joined #parrot
14:30 Coke joined #parrot
14:32 whiteknight bacek: okay, much better
14:33 whiteknight mergin
14:33 dalek parrot: e187a44 | Whiteknight++ | / (16 files):
14:33 dalek parrot: Merge branch 'packfile_wrap'
14:33 dalek parrot: review: https://github.com/parrot/parrot/commit/e187a44654
14:38 ttbot Parrot dbf477e6 i386-linux-thread-multi make error http://tt.taptinder.org/cmdinfo/67253
14:39 bubaflub left #parrot
14:41 moritz wow, rakudo build failures
14:41 nopaste "moritz" at 192.168.1.3 pasted "rakudo build failures on latest parrot" (13 lines) at http://nopaste.snit.ch/40772
14:41 utsl left #parrot
14:42 moritz hm, seems to be related to TT #443.
14:43 ttbot Parrot e187a446 MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/67286
14:43 Andy_ I just was in hll.c last night
14:43 Andy_ that smells like me
14:44 moritz so, what should I replace the call to -#define Parrot_get_ctx_HLL_namespace Parrot_hll_get_ctx_HLL_namespace
14:44 moritz sorry
14:44 jaffa4 to err is human
14:44 moritz what should I use instead of Parrot_get_ctx_HLL_namespace?
14:44 moritz jaffa4: to mispaste is human too
14:44 Andy_ don't know, moritz.  I didn't change any naming.
14:45 Andy_ But I DID make one of the HLL functions return PMCNULL instead of NULL
14:45 jaffa4 not sure what you mean, moritz
14:46 moritz Andy_: ah, found it, never mind
14:46 Andy_ ok, good.
14:47 utsl joined #parrot
14:48 Andy_ Especially since my code note says it was STRINGNULL I was returning. :-)
14:51 whiteknight seriously? Rakudo build failures now?
14:51 moritz whiteknight: seems unrelated to the merge, and I already have a fix
14:51 whiteknight ok
14:52 moritz whiteknight: Andy_++ removed a #define for a long deprecated function which rakudo hasn't bothered to fix
14:52 whiteknight moritz: you are the leading cause of high bloodpressure in parrot devs named "whiteknight"
14:52 whiteknight :)
14:52 whiteknight okay, that's not so bad
14:52 Andy_ moritz: Just doin' what the comment said. :-)
14:53 contingencyplan joined #parrot
14:53 moritz whiteknight: all I'm doing is actually *using* parrot :-)
14:53 Andy_ "I want you to delete me as hard as you can."
14:53 ttbot Parrot dbf477e6 i386-linux-thread-multi make error http://tt.taptinder.org/cmdinfo/67388
14:53 Andy_ Nice, huge swaths of splint errors go away with your merge, whiteknight .
14:53 whiteknight Andy_: is that what the comment actually said?
14:53 Andy_ whiteknight: No.  It just said to delete me after 2.2
14:54 Andy_ So using my awesome powers of numerical comparison, I determined that I should.
14:54 whiteknight okay, that does sound like the kind of ill-advised comment I might leave laying around
14:54 Andy_ Net change on splint warnings: 766 -> 740
14:54 whiteknight okay, nice
14:55 whiteknight I'll have to spend some time playing with splint this coming cycle and see what I can make disappear
14:55 jaffa4 whiteknight:  do you know parrot?
14:55 whiteknight jaffa4: I've heard of it
14:55 moritz Andy_: whatever the comment says, it's nicer to remove such things after a release (as opposed to directly before a release)
14:55 whiteknight jaffa4: you have a question?
14:55 moritz Andy_: no harm done in this case, just a general note
14:55 Andy_ moritz: fair enough.
14:56 moritz Andy_: since there's no canonical way to test if a project uses deprecated parrot features, rakudo generally uses some :-)
14:56 jaffa4 I investigated some problem
14:56 jaffa4 related to parrot on Windows
14:56 jaffa4 parrot did not find perl6
14:56 jaffa4 related files
14:56 whiteknight jaffa4: okay, do you know what the problem is?
14:57 jaffa4 I printed the path used to find the files
14:57 jaffa4 it is not clear where the path comes from
14:57 jaffa4 but it does not come from -I, - L
14:57 jaffa4 switches
14:58 Andy_ One new warning: include/parrot/api.h:25:1: Function Parrot_confess defined more than once    src/exceptions.c:565:1: Previous definition of Parrot_confess
14:58 whiteknight those don't work on windows?
14:58 jaffa4 I did not check if they work
14:59 jaffa4 it is hazy how pbc files are found
14:59 jaffa4 by the parrot
14:59 jaffa4 How should that work anyway?
14:59 whiteknight yes, it is hazy. Should be better
14:59 jaffa4 I cannot tell it is wrong if I do not know how it should work
15:00 jaffa4 I can see it cannot find perl6.pbc file
15:00 jaffa4 and it does not use -I and -L
15:03 whiteknight jaffa4: what directory is the perl6.pbc file in?
15:04 bubaflub joined #parrot
15:06 Coke left #parrot
15:06 Coke joined #parrot
15:08 jaffa4 looking for it
15:10 ttbot Parrot dbf477e6 i386-linux-thread-multi make error http://tt.taptinder.org/cmdinfo/67517
15:12 jaffa4 perl6.pbc in rakudo directory
15:14 jaffa4 the program looks for them in
15:14 giwi left #parrot
15:14 moritz which on is the "rakudo directory"?
15:15 jaffa4 ctrl+v does not work
15:15 jaffa4 left #parrot
15:16 jaffa4 joined #parrot
15:17 moritz weird directory name
15:20 redicaps1 joined #parrot
15:20 redicaps1 left #parrot
15:25 jaffa4 left #parrot
15:26 redicaps left #parrot
15:27 mikehh joined #parrot
15:30 jaffa4 joined #parrot
15:31 jaffa4 http://pastebin.ca/2047814
15:32 jaffa4 as you can see
15:32 jaffa4 D:/src/rakudo/rakudo/parrot/runtim​e/parrot/languages/perl6/perl6.pbc
15:38 whiteknight jaffa4: okay, so it looks like it is searching in that location, from those debug statements you added
15:38 jaffa4 no
15:38 jaffa4 pbc file is in  D:/src/rakudo/rakudo
15:39 jaffa4 whiteknight:
15:39 whiteknight right, that looks like the second-to-last entry that's spit out there
15:39 whiteknight D:/src/rakudo/rakudo/parrot/runtim​e/parrot/languages/perl6/perl6.pbc
15:39 whiteknight D:/src/rakudo/rakudo/parrot/runtim​e/parrot/languages/perl6/perl6.pbc (what you wrote earlier)
15:39 whiteknight those two look the same to me
15:41 dukeleto ~~
15:42 jaffa4 not sure what you mean
15:43 whiteknight jaffa4: I need to figure out if Parrot is searching in the right place, but not finding the file, or if it is not searching in the correct place
15:43 whiteknight jaffa4: it looks to me like it is searching the correct directory, which means there is a problem with Parrot identifying a correct file
15:43 whiteknight maybe I'm seeing something wrong
15:43 jaffa4 as I see it
15:43 jaffa4 the only problem is
15:44 jaffa4 that I cannot add a directory to this path
15:44 jaffa4 the search path
15:44 jaffa4 that is the only prlbem
15:44 cotto_work ~
15:44 whiteknight jaffa4: okay, and -I or -L do not work?
15:44 jaffa4 no
15:44 whiteknight okay, so that's the bug
15:46 jaffa4 there are 4 search paths
15:46 jaffa4 out of 3 I can influence
15:46 dalek parrot: 8393b95 | petdance++ | / (2 files):
15:46 dalek parrot: annotate function pointers in the iterator
15:46 dalek parrot: review: https://github.com/parrot/parrot/commit/8393b9527a
15:46 jaffa4 the forth is
15:46 jaffa4 PARROT_LIB_PATH_LANG
15:47 whiteknight hmmm
15:48 jaffa4 there is no command line switch for that
15:54 dalek parrot: 229a781 | Whiteknight++ | / (2 files):
15:54 dalek parrot: Add experimental ability to modify PARROT_LIB_PATH_LANG with the -L commandline switch.
15:54 dalek parrot: review: https://github.com/parrot/parrot/commit/229a7813ac
15:54 dalek parrot: 46922ae | Whiteknight++ | / (2 files):
15:54 dalek parrot: Merge branch 'master' of github.com:parrot/parrot
15:54 dalek parrot: review: https://github.com/parrot/parrot/commit/46922aec93
15:54 jaffa4 ok
15:54 jaffa4 if I create a path
15:54 whiteknight jaffa4: there is a way now :). git pull and try it out
15:54 whiteknight that commit is experimental, we may pick a better mechanism after the release. For now, use -L
15:54 bacek left #parrot
15:55 ttbot Parrot 8393b952 MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/67870
15:55 jaffa4 it is pretty easy
15:55 jaffa4 12 lines
15:55 jaffa4 to introduce a new switch
15:56 whiteknight jaffa4: yes, it is easy enough to add a new switch. I don't know if that's what we want
15:56 cotto_work which 12 is the question, though
15:56 jaffa4 I found what needs to be changed
15:56 whiteknight adding the switch to the frontend is no big deal. The problem is figuring out which API function to call
15:56 jaffa4 there is no api
15:56 jaffa4 but it can be made quickly
15:57 whiteknight right, do we make a new API function, or modify an existing one?
15:57 whiteknight that's the question, and it's not one we should jump into without some planning and consideration
15:57 jaffa4 the similar ones are in api.c
15:58 jaffa4 see lien 612
15:58 jaffa4 almost copy and paste
15:59 whiteknight right. But that doesn't necessarily mean we want to copy+paste it again
15:59 whiteknight and it certainly doesn't mean that those ones are right in the first place
15:59 whiteknight copy+paste is more the hallmark of bad design, and not ease of duplication
15:59 whiteknight (and I don't mind criticizing those, since I wrote them)
16:02 jaffa4 ok
16:02 jaffa4 think about it
16:03 ttbot Parrot 46922aec MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/67925
16:04 jaffa4 So what are you going to do now?
16:04 jaffa4 whiteknight:
16:09 mj41 joined #parrot
16:09 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#14973) fulltest) at 3_2_0-406-ge187a44
16:09 mikehh Ubuntu 11.04 beta amd64 (g++ --optimize --gc=gms)
16:11 mtk joined #parrot
16:19 dalek parrot: 9ae4a27 | dukeleto++ | NEWS:
16:19 dalek parrot: Add a little meat to NEWS
16:19 dalek parrot: review: https://github.com/parrot/parrot/commit/9ae4a27fb8
16:21 he_ left #parrot
16:24 benabik left #parrot
16:28 dukeleto whiteknight: i see some smolder reports from Win32 that have some failing tests
16:29 bubaflub left #parrot
16:29 dukeleto i think PLATFORM needs to go the way of DEPRECATED.pod -
16:29 dukeleto -> api.yaml
16:29 ttbot Parrot 9ae4a27f MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/68039
16:29 dalek parrot: c992322 | dukeleto++ | PLATFORMS:
16:29 dalek parrot: Update netbsd x86_64 PLATFORM entry
16:29 dukeleto manually updating PLATFORM is very error prone
16:29 dalek parrot: review: https://github.com/parrot/parrot/commit/c9923220e2
16:30 cotto_work dukeleto: what's OS Bridge's policy about speakers paying to attend the conference?
16:30 dukeleto for instance, our smoke reports don't report the OS version for some reason
16:30 dukeleto cotto_work: speakers have always had the conf fee waived, in previous years
16:30 dukeleto cotto_work: how are you getting to LFNW ?
16:31 cotto_work dukeleto: haven't thought about it, but driving seems reasonable
16:34 cotto_work dukeleto: do you know when they decide which talks to accept?
16:35 whiteknight I actually cannot seem to get parrot to build right now on my windows box
16:37 whiteknight we really need more automated building and testing on windows
16:39 dukeleto whiteknight++ # agreed
16:39 whiteknight that's a lesson we seem to remember the day before every release
16:39 dukeleto cotto_work: when OSB decides on talks? Not sure. It is kind of "when they feel like it"
16:40 dukeleto cotto_work: i will probably be taking the train. I was thinking that perhaps we could meet up and continue by train there, and have a Hackathon On Rails
16:40 whiteknight "unresolved external symbol PMCNULL"
16:40 cotto_work punful
16:40 whiteknight great
16:40 dukeleto cotto_work: the train is $60 each way from PDX, i imagine it is much cheaper from Seattle. And you can code on the train :)
16:41 cotto_work this is true
16:41 dukeleto cotto_work: you up for something like that? It would be a great time for an M0 mind-meld
16:41 Andy_ "That's OK, son, you can do it on the boat!"
16:41 cotto_work I'll look into it, though I'll barely have time to think until Wed.
16:42 dukeleto cotto_work: no worries. I will let you know what my plans are, and you can do what ye will :) We will hack on stuff while we are at the conf, no doubt
16:42 dukeleto cotto_work: have you figured out lodging stuff yet?
16:42 cotto_work nope
16:42 * dukeleto neither
16:43 ttbot Parrot c9923220 MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/68098
16:50 jaffa4 What conference?
16:50 cotto_work LinuxFestNW
16:50 cotto_work and later, Open Source Bridge in Portland
16:51 dalek parrot: 8bcd6a7 | Whiteknight++ | frontend/pbc_dump/main.c:
16:51 dalek parrot: fix the build on windows. I don't know why this is an error, the compiler can clearly see the PMC_IS_NULL macro, but claims the PMCNULL variable is not defined. They are defined in the same file
16:51 dalek parrot: review: https://github.com/parrot/parrot/commit/8bcd6a75ea
16:51 dalek parrot: 836fe3b | Whiteknight++ | / (6 files):
16:51 dalek parrot: Merge branch 'master' of github.com:parrot/parrot
16:51 dalek parrot: review: https://github.com/parrot/parrot/commit/836fe3bd30
16:51 mj41 left #parrot
16:52 krunen left #parrot
16:55 dukeleto whiteknight: let me know if you need any specifc platforms tested or if you need other help during the release process
16:55 dukeleto whiteknight: i will attempt to improve NEWS as I see items for it
16:56 whiteknight yeah, I'm going to give NEWS a once-over tonight
16:56 dukeleto whiteknight: also, this might come in handy when writing your release announcement https://gist.github.com/925690
16:56 whiteknight oh nice
16:56 dukeleto whiteknight: that is from the git-extras utility, which you can find on github
16:56 whiteknight I may have to look into that
16:56 * dukeleto finds link
16:57 dukeleto whiteknight: https://github.com/visionmedia/git-extras
16:57 krunen joined #parrot
16:57 whiteknight yeah, I already found it
16:58 dukeleto whiteknight: i added git summary $committish support to it :)
16:58 whiteknight oh nice
16:58 dukeleto whiteknight: it has many nice features. I need to use it more.
16:58 dukeleto whiteknight: anyway, if anything goes pear-shaped before the release, let me know.
16:59 dukeleto whiteknight: looks like we had 18 unique committers to the master branch in the last month
16:59 dukeleto whiteknight: it would be very interesting to track that number each month, as a project vitality statistic
16:59 benabik joined #parrot
17:00 whiteknight don't worry, if things start going crazy between now and tomorrow afternoon, I'll definitely call in a few favors
17:00 jaffa4 left #parrot
17:03 bubaflub joined #parrot
17:06 mikehh rakudo (5ee8c3c) - builds on parrot (3_2_0-406-ge187a44) - make test, make spectest_smolder[(#14978), roast (ea701e8)] PASS
17:06 mikehh Ubuntu 11.04 beta amd64 (g++ --optimize --gc=gms)
17:06 mikehh 27,636 ok, 0 failed, 606 todo, 1,800 skipped and 0 unexpectedly succeeded
17:07 dukeleto mikehh++ # rakudo testing
17:07 darbelo joined #parrot
17:09 dukeleto darbelo: werd up
17:11 mikehh dukeleto: seems to be considerably faster, build (make -j) 3m30.014s, first time in ages I got spectest_smolder (TEST_JOBS=4) to run under 20 min - 19m28.610s
17:12 dukeleto mikehh: very interesting
17:12 dukeleto mikehh: i wonder if that has to do with recent branch merges
17:13 mikehh dukeleto: pretty much so yes - I have been testing all along (not every day though)
17:15 darbelo dukeleto: Hi.
17:15 darbelo How much did I miss in the past few days?
17:19 bubaflub mikehh: bacek mentioned the new GC helped out a lot
17:20 dukeleto darbelo: a few big branches were merged
17:20 dukeleto darbelo: packfile_wrap and imcc_compreg_whatistname
17:20 dukeleto bubaflub: how are finals?
17:21 bubaflub dukeleto: gettin' close - finished up a few classes this weekend
17:21 whiteknight neither of those two branches should have had a markedly beneficial impact on performance
17:21 whiteknight in fact, I would expect small movement the other way
17:21 dukeleto whiteknight: perhaps they have a hidden synergy together
17:21 whiteknight there is some synergy. packfile_wrap is the bugfix branch for imcc_compreg_pmc
17:22 dukeleto bubaflub: what day are you totally done with finals?
17:22 darbelo Packfiles aren't a real bottleneck for HLLs.
17:23 rohit_nsit08 joined #parrot
17:23 darbelo Nor is IMCC once the PIR is done compiling.
17:23 rohit_nsit08 hello #parrot.
17:24 rohit_nsit08 whiteknight: hi
17:24 rohit_nsit08 whiteknight: what is there in 'parrot' pmc?
17:24 mikehh whiteknight: there was a definite improvement (timewise) after the merge of imcc_compreg_pmc (ignoring the segfaults, which seem to have gone)
17:25 whiteknight mikehh: well, that's unexpected
17:25 whiteknight not unwelcome, but unexpected
17:25 whiteknight rohit_nsit08: I don't understand the question. "parrot" pmc?
17:26 rohit_nsit08 var arr  =[1,2] generates PIR code root_new $P1, ['parrot';'ResizablePMCArray']
17:26 rohit_nsit08 i know the root_new and ResizablePMCArray
17:26 rohit_nsit08 what is role of 'parrot'
17:27 rohit_nsit08 ok found it in documentation
17:27 rohit_nsit08 hll namespace
17:29 darbelo Slight misnomer, since parrot isn't a HLL.
17:33 dodathome joined #parrot
17:38 benabik left #parrot
17:41 dukeleto rohit_nsit08: makes sense now?
17:44 whiteknight HLL namespaces are the top-level namespaces. All the default stuff is in the "parrot" namespace
17:44 whiteknight in a Perl6 compiler, all the types would be in the "perl6" HLL
17:44 whiteknight that way compilers don't overwrite things inside Parrot
17:45 ShaneC joined #parrot
17:48 dukeleto mikehh: can you gist your script to test rakudo ? I want to steal it :)
17:49 rohit_nsit08 dukeleto: not exactly, are we defining the new object in a new namespace relative to the default root namespace?
17:50 rohit_nsit08 the format of root_new is "root_new(out PMC, in PMC, in PMC) "
17:50 dukeleto rohit_nsit08: root_new creates a new PMC in the 'parrot' namespace of the type ResizablePMCArray
17:50 dukeleto rohit_nsit08: basically, it creates a new parrot object
17:51 rohit_nsit08 ya the array object in this case
17:51 dukeleto rohit_nsit08: yes, we have many types of array objects, RPA is an array that can hold objects (PMCS) and also be resized
17:52 benabik joined #parrot
17:52 rohit_nsit08 dukeleto: yes
17:52 rohit_nsit08 and we can use inbuilt functions like sort() on the array object
17:52 dukeleto rohit_nsit08: as opposed to a FixedFloatArray, which is a fixed size and can only hold floats
17:52 dukeleto rohit_nsit08: every PMC has an API to interact with it
17:53 dukeleto rohit_nsit08: which are called "vtable functions" or "vtables"
17:53 dukeleto rohit_nsit08: read this for an overview http://leto.net/dukeleto.pl/2011/02/parrot-​embed-grant-update-3-now-with-dragons.html
17:53 rohit_nsit08 I was going through the list of inbuilt functions in javascript which need to be implemented in compiler, most of them i have seen are available in pmc objects
17:53 rohit_nsit08 like sort() for example
17:54 dukeleto rohit_nsit08: you can add any mehtod/function to a PMC, but all of them have a base API that can be added to
17:54 rohit_nsit08 I'm reading the API rightnow . Writing simple programs in winxed and learning from the PIR code generated by it
17:55 dukeleto rohit_nsit08++ # industrious :)
17:55 rohit_nsit08 thanks :-)
17:56 dmalcolm joined #parrot
17:57 dukeleto dmalcolm: welcome
17:58 * dukeleto should probably finish his taxes
17:59 dmalcolm dukeleto: hi!
17:59 dmalcolm go finish your taxes!  :)
18:01 * davidfetter got them in before the deadline date this year
18:02 benabik davidfetter: Me too, although that's because the deadline was late. :-)
18:02 davidfetter heh
18:02 davidfetter i paid a pro to prepare mine, and it was well worth it. that lady is magical
18:03 dukeleto I did the fed, but I have to submit to 2 states because of the whole telecommuting to another state crap. Blarg.
18:03 dukeleto dmalcolm: what brings you to #parrot? I don't reckon I've seen you in here before
18:07 dmalcolm dukeleto: have lurked here for about a year, since allison's presentation of pynie at PyCon US last year (I'm mostly a python person, but am interested in dynamic language runtimes)
18:07 dmalcolm dukeleto: now stop procrastinating and finish your taxes! :-P
18:08 dukeleto dmalcolm: i see. we have a very nice GSoC proposal for Python on Parrot. Exciting times.
18:08 * dukeleto attempts to finish his taxes
18:22 whiteknight taxes are for the weak!
18:25 whiteknight it's easy to say that when my taxes are already done :)
18:39 whiteknight I've learned a valuable lesson this year. I will never do my own taxes again.
18:42 cotto_work I didn't find it to be that much of a pain using an online service.
18:44 whiteknight left #parrot
18:45 whiteknight joined #parrot
18:51 davidfetter sadly, my situation was complicated, but not "hire a bevy of tax attorneys" complicated
18:59 dalek Rosella/test_asserter: bd15f88 | Whiteknight++ | s (4 files):
18:59 dalek Rosella/test_asserter: fixes so test asserter starts building
18:59 dalek Rosella/test_asserter: review: https://github.com/Whiteknig​ht/Rosella/commit/bd15f8839e
19:18 cotto_work dukeleto: did you see "git can't be made consistent"? http://bramcohen.livejournal.com/74462.html
19:19 cotto_work Is there a warning we should put in the git workflow docs about that?
19:19 whiteknight is it something that requires a warning?
19:20 cotto_work seems to be
19:20 whiteknight I didn't think so
19:20 cotto_work though it doesn't match our standard workflow
19:20 whiteknight I read it as non-git-users, complaining about something which they find foreign
19:20 benabik cotto_work: Does that post boil down to "don't do criss-cross merges?"
19:20 cotto_work benabik: that's how I read it.
19:21 cotto_work I can see that very occasionally being an issue.
19:23 whiteknight I still don't think it's an issue worth mentioning. Proper use of a tool is inherent in the tool itself
19:23 whiteknight at least, what constitutes proper use
19:23 cotto_work For rare and unusual cases, I'm fine with "Don't Do That".
19:23 benabik cotto_work: That's a fairly well known axiom on git-list.  General rule of thumb is to merge from master to branch and once branch is merged to master, stop using the branch (or base the branch on the merge in master).
19:25 cotto_work benabik: did I mention I'm excited for your gsoc project?  You seem like you'll work well with bacek.
19:25 Tene I don't see any actual problems described there.  When merging, you should understand what you're merging and how you want it to behave, and in every case, git allows you to get exactly what you want.
19:26 benabik cotto_work: Not directly to me, but I do read the comments on my proposal.  Thanks for your vote of confidence.  :-)
19:26 Tene git won't magically do the right thing in all cases, as that's impossible.  Its heuristic when otherwise-unspecified is right for the majority of normal cases.
19:27 cotto_work Sounds like the consensus is that it's not likely to be a problem.  I'm happy leaving it at that.  Thanks.
19:32 bubaflub left #parrot
19:33 whiteknight blah, every once in a while I am seeing exception backtraces which contain the entire PIR code literal of the input file, instead of the file name
19:34 whiteknight and thats...wrong
19:34 whiteknight I'm already looking forward to debugging that one
19:37 Tene Haha, awesome.
19:37 nopaste "whiteknight" at 192.168.1.3 pasted "backtrace awesomeness" (101 lines) at http://nopaste.snit.ch/40821
19:37 whiteknight the pir code is the file name, and the line number is all sorts of special
19:38 benabik whiteknight: I wonder if the line number is the pointer to the current byte...
19:38 whiteknight benabik: I'm sure it's either a pointer or some random uninitialized value
19:39 dukeleto cotto_work: I would say that there isn't much useful info in that post. It is mostly buzzwords and bad workflows
19:39 dukeleto cotto_work: for instance, I routinely merge master into leto/embed_grant and occasionally merge leto/embed_grant into master, and I have not been set on fire yet
19:40 dukeleto cotto_work: granted, that is a 1 person workflow. If multiple people were doing that, it would be possible to run into problems
19:40 benabik dukeleto: Bram really likes the darcs (history is sum of patches) worldview and appears to be attempting to get git (history is snapshots) to match it.
19:40 dukeleto cotto_work: but multiple people doing that is a bad idea
19:41 dukeleto benabik: the FUD to signal ratio in that post is quite high
19:41 dukeleto benabik: i guess if he likes darcs, he should stick with darcs :)
19:41 benabik dukeleto: That's much my view.
19:42 benabik dukeleto: Although he, like many other, thinks everyone thinks like him.  So he's convinced everyone else has just drank the git kool-aid instead of realizing that some of us prefer the snapshot method.
19:45 benabik Anyway, I gotta go...  stupid school getting in the way of IRC.  ;-)
19:45 dukeleto benabik: yes. If I never had to hear another "git is designed totally wrong because it doesn't act like svn/hg/darcs" conversation again, it would be too soon.
19:45 benabik dukeleto++
19:45 dukeleto benabik: take it easy. I expect awesome things from you soon :)
19:45 benabik left #parrot
19:46 whiteknight seriously. days before GSoC is no time to be acting competent
19:46 whiteknight last thing anybody needs is to set the bar too high for themselves
19:52 bubaflub joined #parrot
20:04 whiteknight heh, I've already figured out that bug
20:05 whiteknight fix should be easy too
20:14 dalek parrot: 6dddd87 | Whiteknight++ | compilers/imcc/ (6 files):
20:14 dalek parrot: fix IMCC_push_parser_state so that backtraces don't contain the full source code of PIR code literals being compiled
20:14 dalek parrot: review: https://github.com/parrot/parrot/commit/6dddd870f4
20:19 dodathome left #parrot
20:19 soh_cah_toa joined #parrot
20:20 whiteknight I would like to find out what that line number is still so far off
20:20 whiteknight the PIR code STRING literal is maybe 100 lines long, and IMCC_push_parser_state resets the line count to 1
20:21 whiteknight so that number makes very very little sense
20:25 bacek joined #parrot
20:26 soh_cah_toa i don't mean to interrupt whatever's being discussed here but i'm curious about the parrot developer summit. i'm filling out my doodle know
20:27 soh_cah_toa how long is the pds? b/c of the fact the it's called a "sumit", i'm guessing there's something special about it
20:27 soh_cah_toa how is it different from an ordinary #parrotsketch meeting?
20:27 jrtayloriv joined #parrot
20:27 sorear it's an extra special #parrotsketch
20:27 sorear a lot of people have awkward schedules
20:27 sorear we know they can only make 4 weekly meetings a year, not 52
20:28 soh_cah_toa is special == long in that statement?
20:28 dukeleto soh_cah_toa: it is kind of like a town hall meeting
20:28 sorear so, we designate 4 meetings as "if you have to make only 4, make these 4"
20:28 dukeleto soh_cah_toa: both special and long :)
20:28 sorear as an indirect consequence, PDSes tend to be longer and much more active
20:29 soh_cah_toa are we talking 1-3 hours? all day?
20:29 whiteknight left #parrot
20:30 soh_cah_toa b/c time wouldn't matter on the weekends. weekdays will require some planning
20:31 sorear I'd say 2-4 hours
20:31 soh_cah_toa oh, that's not bad
20:31 soh_cah_toa so what kind of things go on duringg pds?
20:32 soh_cah_toa as someone new to parrot, what would i be most excited about?
20:38 mj41 joined #parrot
20:39 bubaflub soh_cah_toa: discussion about future direction
20:39 bubaflub soh_cah_toa: setting roadmap goals for upcoming releases
20:40 theory left #parrot
20:40 soh_cah_toa bubaflub: okay
20:40 bubaflub soh_cah_toa: and standard PS stuff - ask questions, talk about new projects, hash out ideas
20:44 soh_cah_toa bubaflub: alright, so it is pretty much just a glorified #parrotsketch
20:44 bubaflub soh_cah_toa: yes; more structured
20:47 soh_cah_toa bubaflub: that's good. i've been wanting to play a more active role in development and be responsible for something. it sounds like pds can help me there
20:48 bubaflub soh_cah_toa: yessir.  they'll probably mention a few roadmap items and you can volunteer to work on one of em
20:48 bubaflub (or more if that's your fancy)
20:49 soh_cah_toa bubaflub: how often do we have pds's?
20:49 bubaflub i think 4 times a year?
20:52 soh_cah_toa bubaflub: alright
20:54 theory joined #parrot
21:00 lucian joined #parrot
21:00 dukeleto PROTIP: All the free tax filing websites don't support telecommuting across state boundaries and having to do multiple state returns. FAIL.
21:01 cognominal left #parrot
21:01 bubaflub dukeleto: you have to file multiple state returns for telecommuting?  i'm not looking forward to that next year...
21:02 dukeleto bubaflub: depends on the state.
21:02 dukeleto bubaflub: i sincerely suggest that you pay someone to do your taxes next year
21:02 cognominal joined #parrot
21:03 theory left #parrot
21:14 utsl dukeleto: I use an online double-entry accounting system, xero.com
21:14 utsl it makes dealing with that shit easy
21:14 utsl then you just need to get an accountant to log in and check it come return time
21:20 dukeleto utsl: sounds useful.
21:20 theory joined #parrot
21:20 utsl the most useful part is the way it just gets fed your transactions and you just mark them as this expense, that expense
21:20 utsl reduces the monthly work to a few clicks
21:21 utsl issues & tracks invoices...
21:21 dukeleto utsl: very nice. I will definitely look into it.
21:21 utsl it has an api too - so services like minutedock.com have cropped up
21:22 utsl which lets you microblog your invoice item
21:22 utsl s
21:22 dukeleto utsl: good lord.
21:22 utsl kind of like twitter, except you get paid
21:22 bacek left #parrot
21:41 ttbot Parrot c9923220 MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/68642
21:43 fperrad left #parrot
21:47 ambs left #parrot
21:53 benabik joined #parrot
21:56 mj41 left #parrot
22:01 benabik left #parrot
22:02 Andy_ left #parrot
22:03 * dukeleto officially hates all tax software and puts a curse upon their houses
22:05 * davidfetter really feels for the people who make tax software, as their deadlines are real and serious
22:05 davidfetter most people don't have dealines /per se/. more like "sick lines" or even "slightly under the weather lines"
22:05 davidfetter ...as in if you don't ship before date X, the sky doesn't actually fall
22:13 rohit_nsit08 left #parrot
22:14 cotto_work April 14th doesn't slip
22:14 cotto_work ;]
22:15 rohit_nsit08 joined #parrot
22:16 mikehh dukeleto: git://gist.github.com/926337.git
22:17 dukeleto mikehh: thanks, I can turn those instructions into a script :)
22:19 soh_cah_toa cotto_work: about the cwd/path issue...you mentioned i should try writing a test. forgive me for not knowing much about testing but what would i need to test?
22:22 cotto_work soh_cah_toa: that would a fairly involved test to write.  Don't feel obligated to worry about it.
22:22 jrtayloriv left #parrot
22:24 soh_cah_toa cotto_work: okay, i'm still getting used to the concept of unit testing. i wish more schools covered it. anyway, should i just wait until wednesday, after 3.3 release, to submit a patch?
22:28 cotto_work soh_cah_toa: you can submit now.  We won't apply it until we're ready (or we can put it in a branch until post-3.3)
22:28 soh_cah_toa cotto_work: okay
22:36 dukeleto soh_cah_toa: if you have questions about testing, let me know
22:37 soh_cah_toa dukeleto: well, the main thing i'm struggling w/ is...what do i test?
22:37 dukeleto soh_cah_toa: this is a very rough draft of an essay about testing that I am writing: https://github.com/leto/writing​/blob/master/misc/lydia_book.md (forgive any typos and such. it was just a dump)
22:37 dukeleto soh_cah_toa: read that essay skeleton and then come ask me questions. Deal?
22:38 dukeleto soh_cah_toa: but the short answer about this specific situation is: change how parrot looks for libraries in the CWD, and then right a test that verifies it "does the right thing"
22:38 dukeleto soh_cah_toa: or just add a test for what happens now
22:41 dukeleto soh_cah_toa: feedback on that writing is greatly appreciated
22:41 dukeleto soh_cah_toa: let me know what doesn't make sense or needs defining
22:41 soh_cah_toa dukeleto: sure
22:43 soh_cah_toa dukeleto: so far, one typo: They are like jetpacks for devevlopers.
22:44 kid51 joined #parrot
22:46 bubaflub left #parrot
22:46 soh_cah_toa dukeleto: you should definitely include some sample code and examples
22:48 dukeleto soh_cah_toa: i am going for very general advice. I don't want it to be specific to any particular programming language or software
22:48 dukeleto soh_cah_toa: but your point is well-taken. I am sure more specific documents would be a good addition
22:48 soh_cah_toa dukeleto: okay
22:48 soh_cah_toa dukeleto: one of the things that's bothering me now is the phrase "units". what defines a unit?
22:48 dukeleto soh_cah_toa: that is touched upon at the end
22:49 dukeleto soh_cah_toa: unit doesn't have a definition. Every situation defines it's own "unit", basically.
22:49 dukeleto soh_cah_toa: it isn't pleasant, but it is the truth.
22:49 dukeleto soh_cah_toa: most often, a "unit" is a function or method.
22:49 dukeleto soh_cah_toa: but it can be different.
22:51 soh_cah_toa dukeleto: this whole testing thing feels weird. apparently, everyone feels that it's a very important part of development. however, no schools teach it and anybody that is good w/ testing has a hard time explaining it b/c it's so abstract
22:53 soh_cah_toa dukeleto: i'm almost embarrased that i've never studied it before b/c it's apparetly very important
22:55 soh_cah_toa dukeleto: it feels like i missed a fundamental part of programming. as if i wrote code for years w/o knowing what a variable was
22:55 dukeleto soh_cah_toa: very few schools teach testing, you are not alone
22:56 dukeleto soh_cah_toa: it is the difference between academia and Getting Shit Done
22:56 dukeleto soh_cah_toa: academics never have to deal with those silly problems like supporting multiple OS's or dependencies changing out from under you
22:57 soh_cah_toa dukeleto: yes! i know!
22:57 soh_cah_toa dukeleto: why the heck not!? is that not what we're preparing for?
22:58 cotto_work Calling code "academic" isn't usually a compliment.
22:59 soh_cah_toa cotto_work: elaborate
23:00 cotto_work That usually means that it's hacky, highly platform-specific and only really proves that something can be done rather than providing a production-quality solution.
23:00 cotto_work not all the time, of course
23:01 cotto_work and to be fair, professors' goals usually involve publishing rather than build and maintaining
23:01 cotto_work *building
23:03 dukeleto cotto_work: academic CS people are mostly mathematicians. They care more about existence and uniquenes (as well as algorithmic complexity) more than maintaining code.
23:03 * dukeleto has to dip out
23:03 dukeleto soh_cah_toa: if you have more feedback about the test essay, please send me an email
23:03 soh_cah_toa dukeleto: sure
23:06 kid51 Question:  I see a lot of commit activity to master today -- much more than in the past few days.  What does that represent?
23:07 kid51 reversion of compreg_imcc merge?
23:08 kid51 Lots of changes to imcc and src/ files.
23:11 davidfetter left #parrot
23:20 whiteknight joined #parrot
23:30 whiteknight I definitely want an abstract parallelization system for Parrot like what is described in that TBB module dukeleto pointed out
23:33 kid51 whiteknight: Which subset of the tests run during 'make test' would be most revealing with respect to --gc=gms?
23:34 whiteknight I don't know. our test suite is notorious for not exercising GC well enough
23:34 whiteknight the ops2c parse tests might be the best for that
23:35 kid51 i.e., sad to say, our *slowest* tests
23:36 cotto_work The best way to exercise gc is to create lots of garbage.  It's hard to write a test to deliberately do that as effectively as something like Rakudo.
23:36 whiteknight that's how you exercise GC, by running the program for a long time
23:37 kid51 cotto_work So, are you saying that Rakudo is excellent for creating garbage? ;-)
23:38 cotto_work In the best way possible.
23:56 whiteknight this really was a boring month, as far as master was concerned
23:57 whiteknight petdance++ going apeshit crazy on our splint warnings like a beast
23:57 whiteknight but it's hard to put that into NEWS in a way that actually makes sense
23:58 whiteknight we should rename parrot to "petdance's big virtual machine with few warnings"
23:59 cotto_work some days it seems that way

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

Parrot | source cross referenced