Perl 6 - the future is here, just unevenly distributed

IRC log for #parrotsketch, 2010-07-27

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

All times shown according to UTC.

Time Nick Message
08:48 robin-gvx joined #parrotsketch
12:21 bluescreen joined #parrotsketch
15:26 mikehh joined #parrotsketch
15:35 bubaflub joined #parrotsketch
15:47 macroron joined #parrotsketch
16:13 eternaleye joined #parrotsketch
16:27 darbelo joined #parrotsketch
16:35 tcurtis joined #parrotsketch
16:59 mikehh joined #parrotsketch
17:11 darbelo joined #parrotsketch
17:26 cotto_work joined #parrotsketch
18:07 NotFound joined #parrotsketch
19:44 atrodo joined #parrotsketch
19:46 khairul joined #parrotsketch
20:00 cotto_work #did:
20:00 cotto_work - met with khairul
20:00 cotto_work - amended support policy
20:00 cotto_work #will do:
20:00 cotto_work - draft a plan and timeline for the git migration
20:00 cotto_work - dig into #1127 (profiling bug due to incorrect packfile debug segments)
20:00 cotto_work #eor
20:00 cotto_work q3q
20:06 Coke Did - - released parrot 2.6.0 - released rakudo 2010.07/Atlanta - minor trac eyeballing - minor partcl-nqp updates.
20:06 Coke Todo: - finally update 'make html' - add namespace support to partcl-nqp (at least as much as is in partcl
20:06 Coke Also: whatever ate those newlines sucks.
20:06 Coke EOR.
20:08 Tene I committed a few lines to nqp-rx repo to implement 'our method' for austin and whiteknight.  I've been waiting on their confirmation before bringing it into the parrot repo.
20:08 Tene EOR
20:08 Tene Looks like I'm starting to recover from burnout just a little bit.
20:13 tcurtis What I did:
20:13 tcurtis * Started implementing Tree::Optimizer.
20:13 tcurtis * Blog post
20:13 tcurtis - http://parrot.org/content/cleaning-and-speeding-optimizations-treeoptimizer
20:13 tcurtis What I will do:
20:13 tcurtis * More Tree::Optimizer stuff.
20:13 tcurtis - Dependencies
20:13 tcurtis - :when/:recursive
20:15 darbelo DONE
20:15 darbelo -   Bought a smallish UPS. No more power outages for me!
20:15 darbelo -   Wasted a *lot* of time refactoring the wrong code.
20:15 darbelo -   Benchmarked stuff. Turns out string speed is mostly irrelevant, compared to PCC and GC in most cases.
20:15 darbelo -   Did a reasonable amount of refactoring on the right code.
20:15 darbelo -   Made the charset struct smaller and a little bit cleaner.
20:15 darbelo -   Made the encoding struct smaller and a little bit cleaner.
20:15 darbelo -   I think there's still stuff that can be moved out of there.
20:15 darbelo TODO
20:15 darbelo -   Keep refactoring the string code.
20:15 darbelo -   See if any of it can be made lazier or smaller.
20:15 darbelo -   Maybe try the "unshared buffers" approach.
20:15 darbelo -   Add more NFG tests.
20:16 darbelo EOR.
20:16 Coke OH HAI Q1Q
20:16 NotFound What I did:
20:16 NotFound -parrot
20:16 NotFound * Just testing
20:16 NotFound -winxed
20:16 NotFound * More float const improvements
20:16 NotFound * Scientific notation
20:16 NotFound * Operatots *= and /=
20:16 NotFound * Allow $ in identifiers
20:16 NotFound * $include_const statement to import pir int constants
20:16 NotFound * Some minor fixes and optimizations
20:16 NotFound * Improved example fly
20:16 NotFound What I will do:
20:16 NotFound No plan
20:16 NotFound EOR
20:19 chromatic joined #parrotsketch
20:21 bubaflub joined #parrotsketch
20:25 mikehh What I did since my last report:
20:25 mikehh * building and testing parrot on amd64/i386, with gcc/g++
20:25 mikehh * some fixes
20:25 mikehh * some work on html_cleanup branch
20:25 mikehh * testing rakudo, partcl-nqp, partcl, pir/PIRATE, winxed and plumage
20:25 mikehh * looking at make splint output to see how I can use it
20:25 mikehh What I intend to do in the next week:
20:25 mikehh * testing and fixing
20:25 mikehh * more preparation as release manager for 2.7.0
20:25 mikehh .eor
20:25 chromatic I'm gathering notes on how to store classes in PBC.  I'll keep on with that and notes on Lorito, and may get to some GC profiling.
20:25 khairul Did:
20:25 khairul - Added more tests
20:25 khairul Will Do:
20:25 khairul - Fix my bugs exposed in the tests.
20:25 khairul - Write more tests
20:25 khairul - Start on user documentation
20:25 khairul EOR
20:28 bubaflub Did:
20:28 bubaflub * Worked on out of directory configuring
20:28 bubaflub Will do:
20:28 bubaflub * Finish configuring / building out of directory
20:28 bubaflub * Start cross-compilation support
20:28 bubaflub EOR
20:31 chromatic Hello, everyone.
20:31 bubaflub hello.
20:31 mikehh hello
20:31 cotto_work hi
20:31 Util # Done: * Win32 rakudo bugs squashed * Helped with Rakudo announcement and CREDITS.
20:31 Util # Plan for next week: * Unstructured catchup * Read all Lorito notes, to be able to speak coherently to Atlanta.pm's concerns.
20:31 Util # Blockers * $WORK * o-o-town half this week
20:31 Util .eor
20:31 Util and, Hi!
20:31 Coke ~~
20:31 darbelo Hola.
20:32 chromatic Let's review last week.  Besides GSoC, I haven't seen many commits for a while.
20:32 mikehh right - not much happened
20:33 chromatic Theories?
20:33 darbelo The magical coding robot has beeen offline since before the relase :)
20:33 bubaflub summer slow down?
20:33 NotFound Hola
20:33 Util Code slush pre-official-supported-release?
20:34 * Util is guilty
20:34 chromatic Not sure what to work on next?
20:34 cotto_work Do you suspect it's more than a natural slowdown?
20:34 NotFound SSD for me
20:34 mikehh and 'parrot wjth a space needs' directory work
20:34 chromatic 2.5 - 2.6 seemed like a slow release.
20:34 Util slow as in performance?
20:35 darbelo Util: commit-wise.
20:35 Util ok
20:35 mikehh nah activity (other that GSOC)
20:35 cotto_work branch merging and pruning might be a good jumpstart
20:36 Coke I did call for slush sooner than usual.
20:36 chromatic I noticed it before the slush.
20:36 Coke naturally occuring slush.
20:37 chromatic Is this something to concern ourselves with, or a natural thing?
20:37 cotto_work It could have soemthing to dow ith Lorito
20:38 chromatic That was my concern as well.
20:38 mikehh possibly, but there are plenty of things that need doing
20:39 cotto_work If that's the case, perhaps the best thing to do is recruit people for Lorito hacking and get the spec hammered out.
20:39 mikehh been looking at the output of make splint, trying to work out where to start
20:39 cotto_work and convince people that hacking on code that will be replaced by Lorito code isn't a waste of time (which I believe it isn't)
20:40 cotto_work s/on code/on C code/
20:40 chromatic We won't replace lots of C code in the near future.
20:40 cotto_work and when we do, we'll look very at the C we're replacing
20:41 chromatic I suppose the question is "How do we motivate people to 1) Bring Lorito into the world 2) Continue our existing system until we can eventually replace it all with Lorito?"
20:41 cotto_work s/very/very carefully/
20:43 mikehh lorito is M0 the rest still needs to work
20:43 cotto_work I can easily picture it being demotivating working on code that has a known expiration date.  Thinking about how to shape that perception is important.
20:43 mikehh and gc is on top of that
20:43 cotto_work mikehh, very much so
20:43 Coke one thing that will help me hack on lorito is a concrete list of tasks.
20:44 Coke ... which i think already partially exists.
20:44 Util chromatic: 2) Assure people of which areas of existing Parrot can be safely improved on, with 0 risk of Lorito clobbering their hard work.
20:44 allison joined #parrotsketch
20:44 chromatic What if I said "If you hack on this part of Parrot, we'll have a much easier time porting it to Lorito?"
20:45 Coke I think I'm the one that raised that concern originally. if no one else has mentioned it, it's probably just me. =-)
20:45 mikehh 1) getting parrot to work properly with a build directory that has spaces in it
20:45 cotto_work Clean and well-encapsulated systems will be much easier to port.
20:46 mikehh 2) improvements in the gc system
20:46 chromatic In the meantime, the faster Parrot runs the better Rakudo can take advantage of it.
20:47 mikehh 3) more test coverage and regression testing
20:48 cotto_work It sounds like a question or two could be added to LoritoDesignQuestions on this topic.
20:48 cotto_work any volunteers who aren't me?
20:48 chromatic I'll take a look.
20:48 NotFound Why on earth is important to build in directories with spaces?
20:49 chromatic If we can't get that right, we don't have shell quoting right.
20:50 NotFound We can't get that right because we use make, and allow to use incompatible versions of make/shell
20:50 mikehh NotFound: it is an allowable option - particularly used in the Windows environment
20:51 NotFound mikehh: try to it with autotools
20:51 NotFound They don't do it right, because of the same reason.
20:51 mikehh NotFound: yeah I know - broken :-}
20:51 NotFound Broken by design.
20:52 chromatic Let's move the technical questions to #parrot.
20:52 chromatic Other thoughts on motivating more development, if that's even possible or desirable?
20:53 Andy joined #parrotsketch
20:53 tcurtis Coke's suggestion for a more concrete task list(or better publicizing an existing one) would be helpful, if possible.
20:54 chromatic Suppose I took a random item off of the Performance Improvement page and made a series of concrete, small tasks.
20:54 chromatic Would people do them?
20:55 Util Anything that the R team says they need from Parrot to achieve part of Rakudo/Perl 6 specs would be motivating for myself, and probably others.
20:55 mikehh We can but try
20:56 Util Performance++
20:56 chromatic http://trac.parrot.org/parrot/wiki/FixingPIRVtableOverrides
20:56 chromatic I'll work on that today.
20:56 chromatic Anything else?
20:57 Coke nice to get the bug stopping the profiling tool from being helpful fixed.
20:58 cotto_work Coke, it's on my todo list
20:58 Util cotto_work++
20:58 chromatic Suggestions for tasks for this week?
20:58 Coke oh. q1q.
21:00 chromatic The crickets suggest moving on to questions.
21:00 chromatic Coke?
21:00 cotto_work need to take off for a meeting.  other questions can go first
21:01 Coke how does PIRC fit into lorito? Is it worth pursuing still?
21:01 Coke (if so, I'll add integrating it as a config option into my queue.)
21:01 Coke if not, we should delete it.
21:02 Util If PIRC is still the annointed successor to IMCC, then quite worth pursuing.
21:02 chromatic Seems like a race between PIRATE and PIRC.
21:02 darbelo Coke: It needs a champion to make it work.
21:02 Util Doh! Forgot about PIRATE
21:02 Tene A new PIR compiler is wanted, yes.
21:03 Tene Whether it's pirc or not isn't important.
21:03 Tene pirc isn't far off of working, IMO
21:03 darbelo OTOH, plobsing seemed to think he could make IMCC suck less at some point.
21:03 Coke Tene: yes, we've heard that, but no one has integrated it yet. =-)
21:03 Util Many improvements to Parrot are not considered, just because IMCC is such a bear to work in.
21:03 Coke right. so we have at least 3 paths forward even sans lorito.
21:04 Coke Util: I have heard the same about pirc. Only one person was championing the code base there.
21:04 whiteknight joined #parrotsketch
21:04 Coke so it sounds like it's not ruled out. I'll keep it on the list, then.
21:04 chromatic Other questions?
21:06 chromatic cotto sent some suggestions for modifications to our deprecation policy to the list.
21:06 chromatic Does anyone have comments?
21:07 Tene I like it.
21:07 Coke That's pretty much what I had in mind when I carped about last minute changes to 2.6
21:07 Coke +1 from me.
21:07 mikehh +1
21:08 chromatic Including the "This never worked, why are we supporting it?" part?
21:08 tcurtis I also like it, though I don't think I really have been around long enough to have any clue about deprecation policy.
21:09 Coke chromatic: yes. the only issue there is we need a concensus on when that is the case.
21:09 Coke but if a sketch full of parrots says so, it must be true.
21:09 whiteknight chromatic: especially that part
21:09 chromatic Then +1 from me
21:09 Util +1
21:10 chromatic Any disagreements?
21:11 chromatic Sounds like rough consensus here then.
21:11 chromatic Other questions?
21:11 Coke reminder: R* release in 2 days.
21:12 chromatic So far, they haven't needed any urgent fixes in weeks.
21:16 chromatic Let's wrap this up then.  Make tasks, everyone!
21:27 bubaflub left #parrotsketch
21:30 cotto_work too late for me?
21:31 cotto_work well crud
22:24 NotFound left #parrotsketch
22:26 atrodo left #parrotsketch
23:25 Util_ joined #parrotsketch
23:25 PerlJam joined #parrotsketch

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