Camelia, the Perl 6 bug

IRC log for #parrot, 2011-10-10

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:05 dukeleto cotto: do you see yourself working on m0 much before the next release?
00:25 l3l1p joined #parrot
02:37 rfw joined #parrot
02:52 dngor joined #parrot
03:33 cotto dukeleto, yes
03:41 dukeleto cotto: what are the priorities? finish the C implementation?
03:42 tadzik hello parrots
03:45 cotto dukeleto, 1) compare x86, x64, arm and possibly mips 2) expand the number of ops in the perl5 implementation so they map ~cleanly to the capabilities of those CPUs 3) make the same changes to the C implementaiton
04:59 l3l1p joined #parrot
05:08 perlite_ joined #parrot
05:37 contingencyplan joined #parrot
05:42 l3l1p joined #parrot
06:37 nbrown joined #parrot
07:16 baest joined #parrot
07:19 SHODAN joined #parrot
08:02 lucian joined #parrot
09:52 lucian Coke: pick Java for what?
09:53 * lucian waves, belatedly
10:09 lateau__ joined #parrot
10:53 dalek rakudo/nom: abc6874 | moritz++ | src/Perl6/Metamodel/C3MRO.pm:
10:53 dalek rakudo/nom: fix c3 MRO, patch by mls++
10:53 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/abc6874fff
11:37 Psyche^ joined #parrot
11:52 dalek nqp: fbe7f0b | mls++ | src/how/NQPClassHOW.pm:
11:52 dalek nqp: backport C3 fix from rakudo
11:52 dalek nqp: review: https://github.com/perl6/nqp/commit/fbe7f0b076
12:17 mtk joined #parrot
12:22 preflex_ joined #parrot
12:27 whiteknight joined #parrot
12:27 whiteknight good morning, #parrot
12:35 Coke joined #parrot
12:52 dalek rakudo/nom: 38907e7 | moritz++ | src/core/Match.pm:
12:52 dalek rakudo/nom: improve Match.gist for failed matches. Also correct Match.perl, masak++
12:52 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/38907e7f56
12:55 jjore joined #parrot
13:32 SHODAN joined #parrot
13:45 PacoLinux joined #parrot
14:33 contingencyplan joined #parrot
14:37 nbrown joined #parrot
14:40 dalek parrot: 08a15fc | gerd++ | / (14 files):
14:40 dalek parrot: change struct name form "Buffer" to "Parrot_Buffer" to avoid name conflicts
14:40 dalek parrot: review: https://github.com/parrot/parrot/commit/08a15fccde
14:53 dukeleto ~~
15:09 mtk joined #parrot
15:16 dmalcolm joined #parrot
15:28 bluescreen joined #parrot
15:28 bluescreen_ joined #parrot
15:57 dukeleto https://github.com/parrot/parrot/wiki/Overview is quite nice
16:08 dukeleto Why does parrot.git contain AIX assembly? https://github.com/parrot/parrot/b​lob/master/src/platform/aix/asm.s
16:08 dukeleto does parrot even build on AIX?
16:12 dalek parrot: 537d8e7 | dukeleto++ | ChangeLog:
16:12 dalek parrot: Add more to the changelog
16:12 dalek parrot: review: https://github.com/parrot/parrot/commit/537d8e71da
16:12 dalek parrot: b79de3a | dukeleto++ | src/platform/aix/asm.s:
16:12 dalek parrot: Remove some jit remnants from our AIX assembly(!)
16:12 dalek parrot: review: https://github.com/parrot/parrot/commit/b79de3ae4c
16:13 whiteknight dukeleto: may have at one time
16:14 whiteknight I can't test it, so I can't say for sure
16:14 dukeleto whiteknight: it would be interesting to see how it fares. Perhaps we can ask the p5p people to test it
16:14 PacoLinux I compiled parrot in aix time ago ..
16:14 dukeleto whiteknight: if AIX is not a supported platform, should we still have aix assembly in our repo?
16:14 dukeleto PacoLinux: interesting!
16:15 PacoLinux (and worked fine)
16:15 dukeleto PacoLinux: how long ago? were you able to run "make smoke" ?
16:16 PacoLinux 1 year and a half, and yes, i think to remember smoke worked
16:16 dukeleto whiteknight: i have been under a rock a bit. Has there been any movement on whether we have a deprecation policy any more?
16:17 dukeleto mls: how does one use the new subprof runcore? Are there docs somewhere?
16:18 dukeleto Also, can somebody tell me exactly what https://github.com/parrot/parrot/b​lob/master/src/platform/ia64/asm.s does ?
16:19 dukeleto that code could benefit from a comment or two
16:19 PacoLinux back in time, I compiled parrot in exotic oses, like irix, solaris(sparc), openvms(alpha) and aix
16:23 dalek parrot: b725c3f | dukeleto++ | ChangeLog:
16:23 dalek parrot: Add note about share/share_ro vtables being gone to the changelog
16:23 dalek parrot: review: https://github.com/parrot/parrot/commit/b725c3f0a7
16:23 dukeleto PacoLinux: do you still have access to some of those machines?
16:37 PacoLinux yes
16:38 dukeleto PacoLinux: it would be very interesting to see if Parrot still compiles and tests ok on aix, if you have spare time
16:39 PacoLinux I will try in other moment
16:43 dukeleto PacoLinux: thanks!
16:43 dukeleto parrot-dev has been eerily quiet lately
16:44 PacoLinux dukeleto: :)
16:46 fperrad joined #parrot
17:16 gbacon joined #parrot
17:40 dalek Heuristic branch merge: pushed 22 commits to rakudo/optimizer by moritz
17:47 cotto ~~
17:47 whiteknight dukeleto: As far as I am concerned, I am acting as if there is no deprecation policy anymore
17:47 whiteknight dukeleto: I
17:48 whiteknight m replacing it with a general attitude of trying to do things with intelligent well-being for the good of Rakudo
17:59 dukeleto whiteknight: ok, i like it. I am trying to think up stuff to write in our changelog
18:01 benabik joined #parrot
18:05 whiteknight ok
18:05 benabik o/ #parrot
18:05 mls dukeleto: simply run parrot with -R subprof
18:06 dukeleto mls: awesome, thanks
18:06 mls currently it writes the profiling output to stdout
18:06 mls when parrot is done, run kcachegrind on the data
18:07 dukeleto mls: do we have this written down somewhere? It sounds perfect for our new github wiki
18:07 mls (there's also the somewhat hidden -R subprof_sub and -R subprof_ops. sub records data on a subroutine level, ops does it on opcode level)
18:07 benabik mls: What does it profile at by default?
18:08 mls the default is subprof_hll, i.e. "annotation block" level
18:08 benabik Annotation block?  By HLL line?
18:09 mls yes
18:09 dalek parrot: 5941717 | dukeleto++ | ChangeLog:
18:09 dalek parrot: Add note about how to use the profile runcore
18:09 dalek parrot: review: https://github.com/parrot/parrot/commit/5941717912
18:09 benabik Neat.
18:13 mls afk -> home
18:25 Coke (new github wiki) what?
18:25 benabik Coke: https://github.com/parrot/parrot/wiki
18:26 benabik soh_cah_toa did a lot of cleaning up of stuff from the old wiki, which is a bit of a mess.
18:26 Coke yes, I know how to /get/ there. when did we start using that?
18:27 whiteknight there wasn't any discussion about it, if that's what you're asking
18:27 Coke (I don't see anything on parrot-dev matching "github wiki" after May 13th, and it's not about the github wiki.
18:27 whiteknight soh_cah_toa started playing with it, it's not an official thing yet
18:28 whiteknight but, if the results are looking good, we might start talking about it more seriously
18:28 Coke Ok. for people who are not following every send in IRC, (and in the logs, since netsplits), pinging things like that to the dev list would probably be a good idea.
18:28 cotto yes
18:28 benabik There was a little bit of talk.  soh_cah_toa said he had been cleaning up wiki-info into his github fork.  Someone else told him "just put it in the parrot one.
18:28 cotto <--
18:29 whiteknight Coke: It's really nothing to talk about yet. It's just an experiment
18:31 Coke please forget I mentioned it.
18:31 benabik Coke: No, you have a good point about mentioning things on -dev.  I tend to forget about it sometimes.  :-/
18:38 whiteknight I wasn't under the impression that it was in a state yet worth mentioning on parrot-dev
18:38 whiteknight I may be outdated
18:39 dngor joined #parrot
18:40 PacoLinux_ joined #parrot
18:45 Coke whiteknight: what's the plan, then: wait until it's ready to replace the existing wiki before mentioning it?
18:46 Coke (wrong order.)
18:46 Coke if it's intended to supplement, not replace, why duplicate effort in the meantime. What about people who find the wiki on github and never see trac? ...
18:56 bluescreen joined #parrot
18:58 alester joined #parrot
18:59 dukeleto Coke: i just sent an email to parrot-dev to spark discussion (as I am wont to do)
19:04 logie joined #parrot
19:11 whiteknight Coke: There is no plan that I am aware of. It's not my project, it's soh_cah_toa's.
19:26 preflex_ joined #parrot
19:45 nine Good evening. I got green_threads working, passing fulltest and Rakudo's spectest. So what's the next step?
19:46 benabik nine: We hold a party for you, I think.  :-D
19:47 Coke nine - like, I can spawn threads and have them do things?
19:47 cotto nine: send a writeup to parrot-dev about what's possible now and what we can do to get a more complete implementation.
19:48 cotto nine++
19:48 nine Coke: tasks :) which will run concurrently to the main task as long as you don't do any blocking I/O
19:48 nine Coke: but it's a start
19:48 Coke nine++!
19:48 Coke I'd love to get that working with auto-threading in rakudo!
19:49 whiteknight nine: if Parrot passes all tests and if Rakudo builds fine, we want to start talking merge and integrate
19:50 nine :)
19:50 nine So I'll write up something tomorrow
19:51 whiteknight that's the substrate we are going to want to build on top of
19:54 Coke nine: you tested "nom" branch on rakudo, yes?
19:54 nine Apart from some small optimizations the next step will probably be to introduce real threads into the mix
19:56 nine And at the same time start thinking about how to use these features in Rakudo and Perl 6. Because from what I've seen here those past months, such things really shouldn't be developed without some serious communications with Rakudo folk
19:56 moritz +Inf to that :-)
20:01 cotto +1
20:02 nine cotto: yes, I tested nom
20:03 nine Coke: yes, I tested nom
20:03 Coke excellent.
20:10 cotto nine: great
20:11 cotto nine: big bonus points if you show some code that can't be done without the changes in green_threads
20:12 nine To be fair: most of what's in green_threads comes from gsoc_threads. I just cleaned it up and fixed a couple of things.
20:12 nine Off to bed now
20:17 whiteknight yeah, we are definitely going to have to explore this branch now, make sure the new stuff is well tested, and talk merge to master
20:18 whiteknight all it provides is a preemptive green threads implementation, and one that is hampered by other design issues in Parrot (nested runloops,etc), so it's a good start but it something we need to build up from
20:19 dukeleto +1 to iterative improvement
20:20 whiteknight what's most important, and I've been saying this for a while, is that we start getting the interface into place so that we can start getting users acclimated to it and start generating feedback
20:21 whiteknight until we get blocking IO sorted out on the branch, it's not going to provide any performance improvements, yet
20:29 rfw joined #parrot
21:28 Coke joined #parrot
21:37 jsut_ joined #parrot
22:27 soh_cah_toa joined #parrot
23:12 whiteknight joined #parrot
23:15 dalek parrot: ea2a265 | jkeenan++ | RESPONSIBLE_PARTIES:
23:15 dalek parrot: Update role.
23:15 dalek parrot: review: https://github.com/parrot/parrot/commit/ea2a265df5
23:23 cotto sad face
23:46 dalek parrot: 0889a4a | jkeenan++ | src/gc/gc_gms.c:
23:46 dalek parrot: [codingstd] Break line so it meets the linelength limitation.
23:46 dalek parrot: review: https://github.com/parrot/parrot/commit/0889a4a7cd
23:46 dalek parrot: 5f8ff5d | jkeenan++ | src/ (8 files):
23:46 dalek parrot: [codingstd] gerd++ reported failures in c_function_docs.t.  These can be
23:46 dalek parrot: cleared up by running 'make headerizer', which will bring function
23:46 dalek parrot: signatures in POD back into line with revised signatures in the
23:46 dalek parrot: functions themselves.
23:46 dalek parrot: review: https://github.com/parrot/parrot/commit/5f8ff5d9d9

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

Parrot | source cross referenced