|
"Tuesday at 20:30 UTC" Set by moderator on 8 July 2010. |
|||
|
00:03
dduncan joined
00:14
dduncan left
04:20
tcurtis joined
04:22
cotto_work joined,
Coke joined,
contingencyplan joined,
Util joined,
PerlJam joined,
japhb joined,
cotto joined,
spinclad joined,
TimToady joined,
ascent joined
06:58
robin-gvx joined
11:07
whiteknight joined
12:38
bluescreen joined
|
|||
| Coke | Did for parrot: Attempted an optimization to speed up rakudo ... which had to be reverted but at least pointed out a missing test. =-) | 13:14 | |
| Will do: hack on partcl; look for more speedups for rakudo; finish htmlcleanup branch; | |||
| . | 13:15 | ||
|
14:07
mikehh joined
14:41
whiteknight joined
15:35
ilbot2 joined
|
|||
| moderator | "Tuesday at 20:30 UTC" | ||
|
16:03
whiteknight joined
16:28
tcurtis joined
17:04
ash_ joined
17:35
NotFound joined
18:28
bluescreen joined
19:18
chromatic joined
19:21
ash_ joined
19:23
Chandon joined
19:43
atrodo joined
19:48
khairul joined
|
|||
| chromatic | I have: sped up Rakudo startup with some targeted Parrot optimizations. | 19:58 | |
| I will: poke at gc_massacre, trying to get it merged. | |||
| Suggest other performance improvements. | |||
| Plead and cajole to get lots of people to help write NQP-rx benchmarks. | |||
| That is all. | |||
| cotto_work | #done: | 19:59 | |
| - met with khairul | |||
| - committed changes to our support policy | |||
| - fixed packfile debug segments #1127) | |||
| #hope to: | |||
| - write or find a trac plugin that'll let us map trac svn changeset urls to git changeset urls (github or other) | |||
| - optimize pprof2cg | |||
| - poke more holes in Lorito's definition, patch them | |||
| #closed: | 20:00 | ||
| - TT #1127 (incorrect filenames in profiling runcore output) | |||
| #eor | |||
| mikehh | What I did since my last report: | ||
| * building and testing parrot on amd64/i386, with gcc/g++ | |||
| * some fixes | 20:01 | ||
| * testing rakudo, partcl-nqp, partcl, pir/PIRATE, winxed and plumage | |||
| What I intend to do in the next week: | |||
| * testing and fixing | |||
| * more preparation as release manager for 2.7.0 | |||
| .eor | |||
| NotFound | What I did: | 20:20 | |
| -parrot | |||
| * More PMC tests added | |||
| * Fixed build in directories with non ascii characters | |||
| * Fixed int fdiv | |||
| * Killed Parrot_PCCINVOKE | |||
| * Replaced remaining usages of Parrot_pcc_invoke_sub_from_c_args with | |||
| Parrot_ext_call | |||
| * Some minor fixes, refactors and nano-optimizations | |||
| -winxed | |||
| * More eye candy in example fly | |||
| What I will do: | |||
| No plan | |||
| EOR | |||
| tcurtis | What I did: | ||
| * Tree::Optimizer work | 20:21 | ||
| - :when and :recursive options to Tree::Optimizer.register | |||
| - :combine option to Tree::Optimizer.run | |||
| - Created PIRATE branch which uses Tree::Optimizer. | |||
| * Added :exact option to Tree::Pattern.ACCEPTS | |||
| * Created branch to implement chromatic's tasklist from FixingPIRVtableOverrides | |||
| - Didn't end up actually working on it. | |||
| * Small amount of gc_massacre profiling. It needs syncing with trunk to benchmark current rakudo and I lack enough knowledge of GC to feel comfortable doing so. | |||
| * Haven't blog posted yet | |||
| * Played with Rakudo * | |||
| What I will do: | |||
| * Blog post | |||
| * More GSoC stuff. | |||
| - Write more optimizations | |||
| - More Tree::Optimizer stuff | |||
| * Suggested 'pencils down' date is the 9th, so will probably focus on cleaning up docs and test after that. | |||
| * Try to work on FixingPIRVtableOverrides stuff. | |||
| EOR | |||
| khairul | did: | 20:26 | |
| -Reworked interface for Instrument | |||
| -Refactored out runcore instruments from Instrument into InstrumentRuncore | |||
| will do: | |||
| -Blog post on changes made above | |||
| -First cut of user guide | |||
| -Fix as much bugs as I can | |||
| eor | |||
|
20:31
darbelo joined
|
|||
| darbelo | DONE | 20:31 | |
| - Finished the encoding - charset code shuffling. | |||
| - Added index and rindex to the binary charset. Part of the TT #1717 fix. | |||
| - Fixed some makefile dependencies in the GSoC branch and trunk. | |||
| - Splinted a bit. Killed nearly 180+ warning with a trivial 2 line diff. | |||
| - Experimented with 'unshared buffers' a bit in my working copy. | |||
| TODO | 20:32 | ||
| - Finish the unshared buffers work. | |||
| EOR. | |||
| NotFound | Hola | ||
| cotto_work | aloha | 20:33 | |
| chromatic | Hello everyone. | ||
| How'd we do last week? | |||
| Coke | meh | 20:34 | |
| chromatic | HTML and tutorial still underway? | 20:35 | |
| cotto_work | some fixes, more tests, more momentum in general | ||
| Coke | I don't think I touched the html_cleanup branch in a week. | ||
| since I missed the release, priority plummeted. | 20:36 | ||
| mikehh | yeah but we want it for the next one | ||
| chromatic | Are there still outstanding tasks somewhere for it? | ||
| tcurtis | The Squaak tutorial? That was mostly updated by the 2.6 release. I did fix some typos/bugs I introduced in the process of updating it, though, this week(Rakudo * attracted at least one set of eyes to Parrot as a compiler target). | 20:37 | |
| mikehh | mainly getting the indexing and stuff | ||
| chromatic | What's the most important thing we could work on in the next week? | 20:39 | |
| in general, not just make html | |||
| Coke | I'll map out what I think is left, as I've left mikehh hanging for a while on that. | ||
| mikehh | trying to get gc_cleanup sorted | ||
| Coke | chromatic: "making rakudo (ok, parrot) faster." | ||
| and lorito. | |||
| and ponies. | 20:40 | ||
| chromatic | I still don't think the big slowdown is Parrot per se. | ||
| Coke | chromatic: no, but profiling rakudo is SLOOOOOW. if we can speed up our profiling tools, that might help. | ||
| mikehh | too much uneeded stuff there | ||
| chromatic | Should we start a task of optimizing the profiling tools? | ||
| cotto_work | I've got that on my todo list already. | 20:41 | |
| (from pre-#ps discussion) | |||
| particle | yes, continue concentrating on the profiling tools, please. | 20:42 | |
| chromatic | I can give you three ideas which should help, back in #parrot. | ||
| What's bacek's availability, does anyone know? | |||
| darbelo | He's been unavailable lately. | 20:43 | |
| I think it was work related. Or singapore related. Or both. | |||
| chromatic | Hm, I've been reluctant to touch gc_massacre and gc_cleanup without him. Pity. | 20:44 | |
| Coke | chromatic: he'll just fix anything you do within a day of his return. | 20:45 | |
| chromatic | The worst I can do is fail to rebase it to trunk. | 20:46 | |
| Other suggestions for what to work on? | 20:47 | ||
| We need to find something for NotFound too. | |||
| NotFound | I'm just doing miscellaneous short things | ||
| chromatic | How do you feel about hashes or vtable overrides? | 20:48 | |
| mikehh | are ther branches that can be merged/worked on? | ||
| there | |||
| NotFound | I'v looked several times at overrides, and never have a good idea. | 20:49 | |
| chromatic | Besides the GC branches, tcurtis started a vtable override branch. | ||
| There's a tasklist on the wiki. | |||
| alternately, looking at a better hash implementation would serve us well even in a Lorito world. | 20:50 | ||
| Coke | +1 | 20:51 | |
| we spend a LOT of time hashing. | |||
| cotto_work | +1 | ||
| NotFound | I can take a look at that | 20:52 | |
| chromatic | +2 | ||
| Let's move on to questions. Anyone have any? | |||
| mikehh | we seem to be having a problem with smolder again | 20:53 | |
| I emailed Michael Peters and he said Sorry, I'm having some problems with that machine. I'm going to have to move smolder some place else. Please bear with me as I find the time to do that. | 20:54 | ||
| shouyld we ask hime to move it to parrot.org? | |||
| or feather or something | 20:55 | ||
| chromatic | It couldn't hurt to ask to see if hosting the Parrot parts elsewhere would work for him. | 20:56 | |
| Other questions? | 20:57 | ||
| tcurtis | Can someone familiar with the C parts of Parrot merge trunk into gc_massacre? I was going to try to do some benchmarking and such last week, and I'd like to be able to test Rakudo HEAD with it. | 20:58 | |
| chromatic | I'll work on that in the next couple of days. | 20:59 | |
| Coke | (smolder) hosting on parrot is fine, but I know the rakudo folks are using it now as well. I say piggyback. | 21:00 | |
| (or we can ask OSU). | 21:01 | ||
| particle | i had osu install smolder for me | ||
| they know how, and it's not hard now that they've done it once | |||
| Coke | ok. have them set us up the smolder, and let us know what the new URL is. we could offer to host the smolder stuff for rakudo too so THEY can piggyback. | 21:03 | |
| (or not. also fine. =-) | |||
|
21:04
Chandn joined
|
|||
| chromatic | Any other questions? | 21:05 | |
| tcurtis | Quick reminder for my fellow GSoC students, the "suggested 'pencils down'" date is next Monday. And the hard pencils down date is the following Monday. | 21:06 | |
| particle | request to osuosl submitted. | 21:07 | |
| thanks for the reminder, tcurtis. | |||
| tcurtis | I have another question, though it's not urgent. Looking at trac, I see some tickets which are fixed but not yet tested and haven't been closed as a result. Is there some standard way to mark such tickets? If not, I'd like to start going through trac looking for such tickets and adding "needs-testing" to their key-words to make them easier to find. Could someone set up a query for that? | 21:16 | |
| chromatic | +1 | 21:18 | |
| NotFound | +1 | ||
|
21:18
ascent joined
|
|||
| mikehh | +1 | 21:19 | |
| tcurtis | Actually, hmm... should I include tickets requesting more testing in this? In which case a "testing" keyword would be more appropriate. Or should "fixed but we need tests to make sure we don't regress" tickets be separate from the "we don't test this thing enough" tickets? | 21:29 | |
| cotto_work | They should be separate. | 21:30 | |
| It's relatively easy to write a test against a known problem. Much less so for an under-tested system. | 21:31 | ||
| tcurtis | True. I'll add a "testing" or similar keyword to those, though, I think. Although better names would be nice. | 21:32 | |
| Maybe needs-regression-tests for "fixed but we need tests to make sure we don't regress"? | 21:33 | ||
| cotto_work | "needs-regression-test" would be good for fixes lacking a test. "testing" would be good for the more general case. +1 | 21:34 | |
| tcurtis starts working on that. | 21:37 | ||
|
21:58
ascent joined
|
|||
| cotto_work | ... and break! | 22:07 | |
|
22:14
Chandon left
22:54
ascent joined
|
|||