Perl 6 - the future is here, just unevenly distributed

IRC log for #parrotsketch, 2010-06-01

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

All times shown according to UTC.

Time Nick Message
04:20 tcurtis joined #parrotsketch
07:39 eternaleye joined #parrotsketch
09:46 mikehh joined #parrotsketch
12:05 bluescreen joined #parrotsketch
13:00 darbelo joined #parrotsketch
17:50 tcurtis joined #parrotsketch
17:51 PerlJam joined #parrotsketch
18:18 bluescreen joined #parrotsketch
18:38 bubaflub joined #parrotsketch
18:43 cotto_work joined #parrotsketch
19:02 khairul joined #parrotsketch
19:04 whiteknight joined #parrotsketch
19:10 whiteknight WHAT I DID:
19:10 whiteknight * Finally did some Parrot work!
19:10 whiteknight * Created ns_func_cleanup branch to cleanup src/global.c and the functions therein.
19:10 whiteknight * Sent an email to the list (warnocked) and requested some information on related tickets (warnocked).
19:10 whiteknight * If anybody has feedback or info on any of TT #1219, #1220, #1221, #1222, #1223, #1224, or #1225, I would REALLY appreciate it
19:10 whiteknight * Created oo_func_cleanup branch. Everything I touch there turns to FAIL. The code is not only extremely ugly but also extremely tightly coupled. I may need to abandon this for now and come back with a fresh game plan later.
19:10 whiteknight * Created a small proof-of-concept PIR subset to C compiler. Starting to explore Lorito in more detail
19:10 whiteknight * Chatting with Chandon about his GSoC project.
19:10 whiteknight * Following along with bacek's GC work, darbelo's GSoC work, and plobsing's opcode work. All very interesting!
19:10 whiteknight WHAT I WILL DO
19:10 whiteknight * Get ns_func_cleanup merged. I would like feedback on the related tickets so I can make fixes in branch if possible. Will merge without fixes otherwise.
19:10 whiteknight WHAT I AM BLOCKING ON
19:10 whiteknight * Not enough hacking time.
19:10 whiteknight EOR
19:17 NotFound joined #parrotsketch
19:21 NotFound What I did:
19:21 NotFound -parrot
19:21 NotFound * Renamed 'unroll' to 'finalize'
19:21 NotFound * Fixed some remaining misuses of str_byte_length/str_length
19:21 NotFound * Cleanup code and pod of some core PMC.
19:21 NotFound * Added a bunch of tests to improve coverage, guided by
19:21 NotFound http://tapir2.ro.vutbr.cz/cover/cover-results/
19:21 NotFound -winxed
19:21 NotFound * Harcoded the value of ITERATE_FROM_START to avoid .include
19:21 NotFound * Avoid using ops recently converted to dynops.
19:21 NotFound * Fixed some problems with new and const noted by plobsing++,
19:21 NotFound but still not a full solution.
19:21 NotFound * Minor fixes and cleanups.
19:21 NotFound What I will do:
19:21 NotFound * More cleanup and test coverage on core PMC.
19:21 NotFound * Anything that crosses my mind.
19:21 NotFound EOR
19:32 plobsing_work joined #parrotsketch
19:38 tcurtis joined #parrotsketch
19:41 bubaflub Did:
19:41 bubaflub * GSoC work for Parrot on RTEMS
19:41 bubaflub - Worked on generating a Configure hints file
19:41 bubaflub - Worked on generating a Configure directive file
19:41 bubaflub Will Do:
19:41 bubaflub * GSoC work:
19:41 bubaflub - Hope to have Parrot configuring and building on RTEMS next week
19:42 Chandon joined #parrotsketch
19:45 Chandon Done:
19:45 Chandon * Exploration of continuations for green threads.
19:45 Chandon Will do:
19:45 Chandon * Look at fitting it in to the scheduler.
19:45 Chandon END
19:50 khairul Done:
19:50 khairul * Learnt enough NQP to reimplement Instrument::Probe in it.
19:50 khairul Will do:
19:50 khairul * Explore replacing op fun ptrs with stubs instead for speed.
19:50 khairul * Look into using the event system
19:50 khairul * Bug squashing tracer.nqp
19:50 khairul END
20:05 plobsing_work What I Did:
20:05 plobsing_work * published not-yet-working deep cloning library on github
20:05 plobsing_work * unleashed ops_massacre. fixed most associated bugs upon prompting.
20:05 plobsing_work * looked into fixing dynops loading (warnocked on list other than good ideas from sorear++)
20:05 plobsing_work * quickly implemented constant unfolding
20:06 plobsing_work What I Plan:
20:06 plobsing_work * fix up residual ops_massacre bugs
20:06 plobsing_work * implement dynops loading bugfix
20:06 plobsing_work * merge constant_unfolding
20:06 plobsing_work * get deep clone library working
20:06 plobsing_work * register allocator???
20:06 plobsing_work Blockers:
20:06 plobsing_work * decision on how we should approach dynops problem (RFC on list has ideas, but no real resolution)
20:06 plobsing_work EOR
20:12 Util # Done:
20:12 Util * Poked Coverity at a different email address, asking to restart scans of Parrot.
20:12 Util = David Maxwell did not respond to the March email.
20:12 Util * Started new presentation: "Top 10 things that are/aren't the same in Perl 6"
20:12 Util = As mused on here: http://irclog.perlgeek.de/parr​otsketch/2010-02-09#i_1969154
20:12 Util # Plan to do:
20:12 Util * Finish Top10 presentation.
20:12 Util * Assist with Perl6book.
20:12 Util * More on TT#1302 (PIR todo() is frequently misused).
20:12 Util # Blockers:
20:12 Util * $WORK
20:12 Util .end
20:13 tcurtis What I did:
20:13 tcurtis * Fixed an issue where my pastcompare.pir helper for testing was getting deleted by make clean.
20:13 tcurtis * Added PAST::Walker::Dynamic and PAST::Transformer::Dynamic.
20:13 tcurtis * Added tests for PAST::Walker::Dynamic and PAST::Transformer::Dynamic.
20:13 tcurtis * Switched PAST::Walker and subclasses to use P6metaclass for more convenient NQP usage.
20:13 tcurtis * Met with pmichaud to discuss PAST optimization.
20:13 tcurtis * Thought about syntax for PAST optimization.
20:13 tcurtis - Sent an email to parrot-dev about it.
20:13 tcurtis * Also asked for any anyone interested in my project to contact me so that I can contact them specifically for feedback when making decisions about what to do.
20:13 tcurtis - pmichaud was the only one who responded. Anyone else interested?
20:13 tcurtis * Posted blog post
20:13 tcurtis - http://parrot.org/content/pasttransfo​rmer-and-foundation-past-optimization
20:13 tcurtis * Started(not even commited yet implementing PAST::Pattern
20:13 tcurtis - Can build whatever syntax idea ends up sounding best on top of it.
20:13 tcurtis What I will do:
20:13 tcurtis * Continue implementing PAST::Pattern.
20:13 tcurtis * Write tests for PAST::Pattern.
20:13 tcurtis * Miss next week's #ps.
20:13 tcurtis - Will still make a report Tuesday before I leave.
20:13 tcurtis * If time allows, do one of two things:
20:13 tcurtis - Start writing example optimizations with PAST::Pattern or
20:14 tcurtis - Start trying out syntaxes for optimization.
20:14 tcurtis EOR
20:16 cotto_work #did:
20:16 cotto_work - helped khairul figure out some stuff
20:16 cotto_work - got feedback on LoritoRoadmap from plobsing and whiteknight
20:16 cotto_work - feedback from everyone (this means you) is welcome
20:17 cotto_work - made a rough meta-plan for Lorito hacking and design
20:17 cotto_work #will do:
20:17 cotto_work - post Lorito meta-plan to parrot-dev
20:17 cotto_work - get cracking on the git/trac test site
20:17 cotto_work - maybe write some code or something
20:17 cotto_work #eor
20:18 bacek joined #parrotsketch
20:19 Coke joined #parrotsketch
20:21 Coke Done: - not much, holiday weekend. - verified that plobsing++'s fix to partcl worked. - started chat on parrot-dev about dynops
20:22 Coke WillDo: - get partcl working again. - keep on top of any parrot issues that impact Rakudo.
20:22 Coke RANT: irssi ate my newlines.
20:22 Coke EOR
20:24 bacek Did:
20:24 bacek * Start GCMassacre. Going reasonably well. Able to build corevm, pass number of coretest. PMC allocations are faster then trunk. Overall performance little bit slower because of PMC Attributes, STRING header allocations aren't properly implemented (yet)
20:24 bacek Todo:
20:24 bacek * Finish GCMassacre
20:24 bacek Blockers:
20:24 bacek * 'constant' PMCs are stand in way of new GC.
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 * various fixes
20:25 mikehh * branch testing and some fixes
20:25 mikehh What I intend to do in the next week:
20:25 mikehh * testing and fixing
20:25 mikehh .eor
20:26 darbelo DONE
20:26 darbelo -    More NFG work. Basic grapheme table functionality has arrived.
20:26 darbelo -    Mostly within the bounds of the initial draft schedule.
20:26 darbelo -    I've dicovered I suck at API design.
20:26 darbelo -    I'm just adding stuff as the need arises in a feeble attempt to avoid overdesign and bloat.
20:26 darbelo TODO
20:26 darbelo -    Blog post about this week progress. Will be done in a few minutes.
20:26 darbelo -    Turn the Grapheme table prototype into something a bit more usable.
20:26 darbelo -    De-uglify code.
20:26 darbelo EOR.
20:32 cotto_work anyone care to get this party started
20:33 Coke I can, I guess.
20:33 Coke Any last minute report-pastes?
20:35 Coke ... Anyone have any feedback on this week's goals? (digging...)
20:35 Coke looks like GC was the big goal. bacek, anything to add to your report?
20:36 bacek Coke, nope. I hate "constant" PMCs...
20:36 darbelo zombies--
20:36 mikehh major probslems with make corevm/make coretest this week
20:37 mikehh problems
20:37 bacek I started gc_massacre_no_constant branch to kill "constants"
20:37 Coke bacek: ok. are you stuck on them, or just delayed?
20:37 cotto_work tons of unintended consequences from ops_massacre
20:37 plobsing_work yeah, sorry about that. I try to fix them as I become aware of them
20:37 cotto_work plobsing_work: and doing a good job of it
20:37 Coke I think the biggest thing that would have helped there is if we had done a 'make coretest' before the mergeback. we have to make sure that keeps working, even if developers don't happen to run it every day.
20:37 mikehh yes - got most working then bacek made a change that blew it up again
20:38 Coke was bacek's the "don't build dynops for core"? that was actually the right patch, IMO.
20:38 NotFound plobsing_work: no your fault, the problem was already present, just it doesn't manifested so clearly.
20:38 cotto_work what NotFound said
20:40 plobsing_work q1q
20:40 mikehh at the moment there are two subtests failing make test but a whole bunch failing make corevm/make coretest
20:42 Coke mikehh: I'm only getting 2 on the corevm atm - if you have more, feel free to hijack my ticket.
20:42 Coke do devs here have an opinion on keeping dynops non-core?
20:42 plobsing_work +1 to non-core
20:42 Coke metoo.
20:43 NotFound +1
20:43 mikehh Coke: let me check - I last ran the tests a few hours ago
20:43 Coke In that case, I think we need to at the very least address notfound's issue with stderr.
20:43 Coke resurrecting printerr would probably be the simplest solution, or blessing that experimental method (which i don't like, as it is using 0,1,2...)
20:44 bacek +1 to move printerr back to ops.
20:44 plobsing_work what's the problem? hardcoded stdhandle indices? we can put those in a .pasm include
20:44 cotto_work +1
20:44 NotFound No preference, whatever works.
20:44 Coke once we decide on that, we can work on fixing/moving all the tests.
20:45 Util +0, as long as all dynops get tested under `make test`
20:45 Coke Util: aye, that's for sure.
20:46 NotFound We can add one or several methods in the FileHandle PMC, if fits better here than in the interpreter.
20:46 Coke alright. I'll sum up to the mailing list and get some more feedback.
20:46 Coke I think the rest of the ops are arguable, but printerr seems to be a pain point.
20:46 Coke plobsing_work: your q?
20:47 NotFound BTW if we put a non-op way on core it will be the prefered way and then we can kill printerr.
20:48 plobsing_work can I get a decision on how to resolve the dynops load order issue (posted to the list as RFC)?
20:48 mikehh Coke 45 tests fail and a lot not even run
20:48 NotFound plobsing_work: ENOARCHITECT
20:48 plobsing_work it is blocking resolution of the major dynops breakage
20:49 Coke plobsing_work: I'll try to review it, but will likely just poke chromatic & allison et al.
20:50 NotFound I think we must choose the faster implementable and testable way, and reevaluate the decision after Rakudo*
20:52 Coke any other non-dynop issues?
20:52 Coke I think we should make that our priority for the next week.
20:52 NotFound +1
20:52 mikehh +1
20:53 cotto_work sounds like what's needed is feedback from chromatic and allison.  I don't mind diving in to help hack if needed.
20:53 Coke ISTR allison is booked with school this week, so we'll try to route around that.
20:54 Coke anything else?
20:55 bacek I would like to get some help on gc_massacre_* branches.
20:55 cotto_work what kind of help?
20:55 cotto_work eyeballs, bug fixing, testing, other?
20:56 Util Coverity got back to me just now; scans to resume tomorrow.
20:56 bacek all of them
20:56 mikehh bacek: working on bits of it
20:56 bacek mikehh, great!
20:56 cotto_work good thing we have a wiki page for the branch
20:57 bacek _no_constants branch is higher priority for now.
20:59 Coke bacek: ok. if you keep the wiki page up to date, sounds like you'll have some volunteers.
20:59 Coke Util: excellent.
21:00 Coke so, 2 priorities - gc massacre branch(es), and fixing dynop issues.
21:00 mikehh sounds good to me
21:00 Coke (but not in that order =-)
21:00 Coke alright. I'm ready to call it, unless anyone has something else?
21:01 Coke ... bueller?
21:02 tcurtis Does anyone other than pmichaud and sorear have any particular interest in my GSoC project? Particularly, if you think you'll actually end up using it, I'd like to know so I can contact you specifically when I need feedback about design decisions.
21:02 bacek tcurtis, I'm.
21:03 Coke tcurtis: I have some interest, primarily as a consumer of PCT.
21:03 mikehh tcurtis: yes - happy to help with testing issues
21:05 Coke tcurtis: if you're looking for more interested parties, parrot-dev will have a higher hit ratio than #ps. (esp for those who can't be here.) - or a blog post works, too.
21:06 NotFound I don't use PCT but always enjoy discusions about syntax ;)
21:06 Tene I have interest.
21:06 tcurtis Coke: I asked on parrot-dev last week. Lack of response was the reason I brought it up here.
21:06 Tene (although I'm not actually here)
21:06 plobsing_work left #parrotsketch
21:08 tcurtis I've added all of you to my list of people to contact. Anyone else?
21:12 Coke tcurtis: it may be no. pmichaud was the primary developer on PCT .
21:12 Coke I'll followup on list to drum up more replies.
21:13 Coke alrightee. Let's call it a week. See everyone on list and next week.
21:14 tcurtis joined #parrotsketch
21:14 bubaflub left #parrotsketch
21:19 khairul left #parrotsketch
21:20 NotFound left #parrotsketch
21:23 whiteknight joined #parrotsketch
21:27 Chandon left #parrotsketch
21:33 tcurtis joined #parrotsketch

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