Perl 6 - the future is here, just unevenly distributed

IRC log for #parrotsketch, 2009-06-30

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

All times shown according to UTC.

Time Nick Message
01:28 japhb joined #parrotsketch
02:06 amuck_ joined #parrotsketch
03:09 amuck joined #parrotsketch
04:13 particle1 joined #parrotsketch
04:26 eternaleye joined #parrotsketch
05:00 amuck joined #parrotsketch
05:03 amuck_ joined #parrotsketch
10:31 kid51 joined #parrotsketch
10:45 kid51 kid51's report
10:45 kid51 * Have returned to NYC and $job (hence, only available for pre-posting)
10:45 kid51 * I have 3 patches lined up for application after the July 21 deprecation point:
10:45 kid51 * https://trac.parrot.org/parrot/ticket/108
10:45 kid51 * https://trac.parrot.org/parrot/ticket/440
10:45 kid51 * https://trac.parrot.org/parrot/ticket/490
10:45 kid51 * There are two other tickets which I would really like to nail down before the July release, but will need active feedback from certain core committers:
10:45 kid51 * https://trac.parrot.org/parrot/ticket/786:  auto::gettext broken on Darwin
10:45 kid51 * I now know that this is more than just extraneous warnings during Configure.pl; gettext is not being detected
10:45 kid51 * need feedback from particle (whose commit last year caused the breakage) on what original reason for change was and how we can work around it, given that code in that region of config/auto/gettext.pm has undergone subsequent refactoring
10:45 kid51 * https://trac.parrot.org/parrot/ticket/797:  eliminate need to expand Makefile variables in parrot_config and other extrrnal programs
10:45 kid51 * Am approaching this a little bit at a time; see TT.  But suspect that the right thing to do is to banish ALL conditional determination of configuration values in Makefile, as the results of such determinations are not available (via lib/Parrot/Config/Generated.pm, config_lib.pasm) to other programs.  Will be seeking feedback from jorwitz, allison and doughera.
10:45 kid51 * Am eagerly awaiting pmichaud's post on 'make install' issues, promised at YAPC, as that will affect what I and wayland have to work on
10:45 kid51 EOR
13:01 Whiteknight joined #parrotsketch
13:36 pmichaud joined #parrotsketch
14:38 davidfetter joined #parrotsketch
15:41 rdice joined #parrotsketch
16:03 einstein joined #parrotsketch
16:18 cotto # What I did:
16:18 cotto * YAPC! - many productive discussions, all the important ones being captured elsewhere
16:18 cotto * lots of pmc_pct hacking - class_init looks like it's being generated correctly now
16:18 cotto - also got some cleanup work done while stuck at the airport, PIT++ for having free wireless
16:18 cotto - however...
16:18 cotto * bacek had an interesting suggestion that could make pmcc obselete
16:18 cotto - instead of working on pmcc, get an ops compiler working
16:18 cotto - once that's in place, we'll have most of the tools we'll need to implement PMCs in pure PIR/nqp/etc without needing a specialized compiler
16:18 cotto - for now, I've put pmcc on hold
16:18 cotto - the ops roadmap is:
16:18 cotto + Dig through the ops2c code, figure out what all the compiler will need to do. (YOU ARE HERE)
16:18 cotto + Make a pct-based compiler for ops that emits the same C code that ops2c does now.
16:18 cotto + Specify enough of L1 to write the ops in it (either directly or in something that compiles to L1).
16:18 cotto + Make PCT capable of transforming L1 into C.
16:18 cotto + Extend the ops compiler to accept other languages (L1, PIR, nqp, bf, INTERCAL, etc).
16:19 cotto + Reimplement ops in terms of L1 (or whatever langauge makes sense).
16:19 cotto * I started trying to rewrite Hash in nqp, started a wiki page to document what's needed to make it work.
16:19 cotto - There are some issues which will be tricky to solve (typing, struct access, struct definition).
16:19 cotto - Some things may be easier in nqp (methods on STRINGs, all memory management going through the GC, string/int conversion).
16:19 cotto - I'm not convinced that nqp is the best language for this job.
16:19 cotto # What I hope to do and how many tuits I expect to have:
16:19 cotto * I'll start figuring out the subtleties of ops2c and friends.
16:19 cotto * I'll also be looking at callgrind's output formatting with the goal of making Parrot spit out something similar.
16:19 cotto * My tuit outlook is good and I expect to get some stub code committed before the next #ps.
16:19 cotto # What could block my progress:
16:19 cotto * There are no external blockers for what I expect to accomplish.
16:20 cotto eor
16:40 jonathan joined #parrotsketch
16:49 jonathan Pre-pasting report...
16:49 jonathan This week...
16:49 jonathan * Got back from holiday
16:49 jonathan * Mostly caught up on reading the Perl 6 spec diffs from while I was away
16:49 jonathan * Started refactoring traits handling to be more in line with latest spec changes, to help more towards my ongoing Hague Grant.
16:49 jonathan * Had some useful discussions with Larry on traits and MMD, which resulted in some clarifications to S12, some of which lay down in the spec stuff I've been assuming up until now. I had a happy (but now have gotta go work out how to deal with some of the things that it's clarified need to happen).
16:50 jonathan * subset types without a where now work, and you can specify literals in signatures and have those work too
16:50 jonathan * Various other fixes and closing up tickets that are already resolved...between pmichaud++ and I we now have the queue back under 400 tickets...which is still a load.
16:50 jonathan In the coming week...
16:50 jonathan * Much more work on traits
16:50 jonathan * Continuing to work on more bits of D1 of my dispatch refactoring grant
16:50 jonathan * Probably a Rakudo day somewhere, which given the size of the RT queue will probably go on trying to reduce that a bit
16:50 jonathan * Need to work out what talks to submit at YAPC::Asia, Italian Perl Workshop and possibly Baltic Perl Workshop too.
16:50 jonathan Blockers
16:50 jonathan * Waiting on PCC refactors
16:50 jonathan * Need to discuss Parrot's MultiSub PMC a little.
16:50 jonathan .end
16:50 jonathan (will write to list aobut the Parrot MultiSub stuff though, it's probably better done there).
16:50 Topic for #parrotsketchis now New proposed #parrotsketch protocol: https://trac.parrot.org/parrot/wiki/ProposedParrotsketchProtocol | Please prepost reports by 1800 UTC.
16:50 smash joined #parrotsketch
16:51 NotFound joined #parrotsketch
16:59 moritz joined #parrotsketch
16:59 Topic for #parrotsketchis now New proposed #parrotsketch protocol: https://trac.parrot.org/parrot/wiki/ProposedParrotsketchProtocol | Please prepost reports by 1800 UTC. | Reports can be seen on http://irclog.perlgeek.de/parrotsketch/today
17:01 Topic for #parrotsketchis now New proposed #parrotsketch protocol: https://trac.parrot.org/parrot/wiki/ProposedParrotsketchProtocol | Please prepost reports by 1800 UTC. | Logs: http://tinyurl.com/parrotsketch
17:09 Coke joined #parrotsketch
17:10 mikehh joined #parrotsketch
17:12 Whiteknight #Last 2 Weeks
17:12 Whiteknight - YAPC::NA
17:12 Whiteknight - Found some problems with threading, and helped Infinoid++ fix some related leaks.
17:12 Whiteknight - Have met some new hackers interested in working on threading, so that may become a group project soon
17:12 Whiteknight - Working on IO refactors in the io_cleanups branch
17:12 Whiteknight - io_cleanups is stalled temporarily while we better integrate the new Pipe type
17:12 Whiteknight - Coke++ started a branch to convert Contexts to a PMC type
17:12 Whiteknight - I'm working on that now, give 50% chance that the branch will succeed.
17:12 Whiteknight - Finding lots of places where contexts need to be cleaned up and better encapsulated.
17:12 Whiteknight - Planning out a new GC core based on the VCGC design. Been playing with prototype code locally for that
17:12 Whiteknight - Laying final ground work for AIO, which I hope to start soon (pending io_cleanups branch)
17:12 Whiteknight - Planning to fix the PMC subclassability problem, thanks to Allison++. See TT #789
17:12 Whiteknight - Lots of talking and speculating about L1, becoming more convinced about it's benefits
17:12 Whiteknight - Blogging about and answering questions about L1
17:13 Whiteknight - Blogging like a madman to try and increase interest among new users post-YAPC. Have gotten good feedback
17:13 Whiteknight - Meeting lots of new hackers, answering questions, and applying patches.
17:13 Whiteknight #This Week
17:13 Whiteknight - Continuing context PMC work, will know by midweek if branch will succeed as-is or if we need an intermediate cleanup step first
17:13 Whiteknight - Getting io_cleanups branch fixed and merged to trunk.
17:13 Whiteknight - Starting on AIO
17:13 Whiteknight #Blocking on
17:13 Whiteknight - Nothing
17:13 Whiteknight EOR
17:24 Coke report coke {
17:24 Coke done {
17:24 Coke item http://partcl.blogspot.com/2009/06/improved-interactive-mode.html
17:24 Coke item started context_pmc branch (to convert contexts over to PMCs)
17:24 Coke }
17:24 Coke todo {
17:24 Coke item work with whiteknight on context_pmc branch.
17:24 Coke item
17:24 Coke }
17:24 Coke blockers {
17:24 Coke item anything marked as lang=tcl in trac.
17:24 Coke item anything at http://code.google.com/p/partcl/wiki/ParrotIssues
17:24 Coke item the fact that PIR PMCs cannot do everything C PMCs can.
17:25 Coke }
17:25 Coke }
17:28 moritz - parrot: no work done
17:28 moritz - rakudo
17:28 moritz + implemented basics of series (infix:<...>) and eqv operators
17:29 moritz + usual testing stuff
17:29 moritz + blogging about Perl 6 and Rakudo
17:29 moritz :wq
17:29 moritz oh, one thing I forgot: I compiled parrot with clang+llvm, it now works --without-gmp
17:29 moritz :wq!
17:43 NotFound * Added function mem_sys_strdup and replace usages of str_dup with it.
17:43 NotFound * Fixed some imcc segfaults.
17:43 NotFound * Fixed some imcc memory leaks with filenames.
17:43 NotFound * Working in TT #767, multiple .local with same name. Emitting a imcc warining for a now.
17:43 NotFound EOR
17:43 allison joined #parrotsketch
17:53 Util joined #parrotsketch
17:54 Util # Done:
17:54 Util * Attended Day 1 of the Parrot VMW, and whispered about Perl 6 shinyness to my neighbors.
17:54 Util * Found that Parrot's substr does not implement negative length (Perl 5 does). Discussed with Whiteknight.
17:55 Util * Participated in Parrot Implementors BOF.
17:55 Util * Discussed calling conventions issues with pmichaud and chromatic.
17:55 Util * Recovered from YAPC.
17:55 Util # Plan for the week (or so):
17:55 Util * Write ticket for negative-length-substr.
17:55 Util * Study up on Parrot's calling conventions, in case I can be of help.
17:55 Util * Catch up writing tickets for issues found 3 weeks ago.
17:55 Util * Close pbc_to_exe ticket, with notes on how to patch further when future compilers need similar coddling.
17:55 Util - Well, chromatic closed it. I will update and re-close.
17:55 Util * Dust off my tinkertoy P5/PPI-based P5-to-P6 translator, add pointers to other implementations, and give it a home on GitHub.
17:55 Util # Blockers:
17:55 Util * perl6-examples/shootout stalled, due to branch-of-a-fork issues. Low priority, unlikely this week.
17:55 jhorwitz joined #parrotsketch
17:56 Util * Minimal tuits through the weekend.
17:56 Util .end
17:57 jhorwitz last week: at YAPC.  helped w/ some installs of mod_parrot/mod_perl6.
17:57 jhorwitz refactored mod_parrot configuration system to be less kludgy
17:57 jhorwitz this week: working on PDD10 for the foreseeable future
17:58 jhorwitz EOR
18:01 pmichaud Pm reports:
18:01 pmichaud What I did (since 06-16-2009):
18:01 pmichaud * Released Rakudo #18 ("Pittsburgh")
18:01 pmichaud * Rakudo now passing 11,515 spectests (-20 since two weeks ago)
18:01 pmichaud * Rakudo is failing 78 tests (mostly related to -G bugs)
18:01 pmichaud * Just got back yesterday from PVMW+YAPC+vacation
18:01 pmichaud * Parrot Virtual Machine Workshop was _very_ successful
18:01 pmichaud * Full blog posts later tonight
18:01 pmichaud * Fixed a bug in unicode upcase when ICU not present
18:01 pmichaud * Updated Rakudo to be able to build from non-svn Parrot builds
18:01 pmichaud * Rakudo can now define operators in Perl 6 (settings)
18:01 pmichaud What I'm doing this week:
18:01 pmichaud * Blogging and catching up on PVMW+YAPC tasks
18:01 pmichaud * Rakudo day 1 (today):  cleaning up Rakudo tickets in RT queue
18:01 pmichaud * Rakudo day 2 (???)  :  container rw handling, BEGIN blocks
18:02 pmichaud * Provoking and responding Parrot install and build tree issues
18:02 pmichaud * Solidifying funding for grants, travel
18:02 pmichaud What I'm blocking on:
18:02 pmichaud * Parrot -G bugs / memory corruption
18:02 pmichaud * Cleaner mechanism for building from dynops/dynpmcs
18:02 pmichaud * Installed pbc_to_exe still relies on build tree
18:02 pmichaud * Various perl6-tagged tickets in Trac
18:02 pmichaud q1q
18:02 pmichaud eor
18:04 amuck_ joined #parrotsketch
18:07 allison - Finished editing and typesetting the book. Submitted it to the printer. Checked the updated book back into trunk.
18:07 allison - Started working on eliminating the final few build-directory dependencies in the compiler utilities.
18:07 allison EOR
18:13 darbelo joined #parrotsketch
18:17 darbelo .report decnum-dynpmcs :GSoC
18:17 darbelo * No longer sick!
18:17 darbelo * Put the PCT test converter on hold.
18:17 darbelo * Bashed together a perl script to avoid getting delayed.
18:17 darbelo * We have tests! We fail tests! Started work on getting them to pass.
18:18 darbelo * We are apparently miss-rounding on some cases. Need to investigate more.
18:18 darbelo * Changed all numeric getters/setters to return/accpet INTVAL.
18:18 darbelo * Re-thinking the way we throw exceptions. It might need some corrective surgery.
18:18 darbelo * I am forgeting something else for sure.
18:18 darbelo .end
18:20 chromatic joined #parrotsketch
18:20 chromatic I fixed some bugs.
18:20 chromatic I closed some tickets.
18:20 chromatic We had a nice retrospective at YAPC.
18:21 chromatic cotto and I discussed how to make a profiling core.
18:21 chromatic This week I'll work on that, specifically refactoring runcores to make them easier to plug and play.
18:21 chromatic I'm blocking on fixing the Context memory leaks until some sort of context/invocation cleanup lands in trunk.
18:24 amuck_ joined #parrotsketch
18:30 japhb Done: Start of work on OpenGL::Math, to make a useful case study for more HLL-to-parrot calling convention issues.  Todo: Continue that.  Blocking on: Found problem with sub-namespaces (so when you 'use OpenGL::Math:from<parrot>;', you aren't able to instantiate OpenGL::Math::Vec4 (and adding double colon at the front (::OpenGL::Math::Vec4) just causes fail).  Tene knows about this and is thinking about it.
18:31 cotto HELLO
18:31 moritz oh hai
18:31 japhb o/
18:31 allison hello
18:31 jhorwitz hello
18:31 smash greetings
18:32 darbelo hola
18:32 Util hello
18:32 pmichaud say "\c[104,101,108,108,111]";
18:34 NotFound hola
18:34 jonathan hi
18:34 jhorwitz pmichaud++ # hello golf
18:34 jonathan golf? It was more chars than anyone elses!
18:34 cotto more like flog
18:34 jhorwitz true
18:34 Whiteknight joined #parrotsketch
18:35 moritz who's master of ceremonies today?
18:35 allison I guess the first point of order is the changed procedure.
18:35 allison It looks like everyone preposted their reports.
18:35 allison Did we miss anyone?
18:36 chromatic Sounds like no.
18:36 chromatic What's our vision for the 1.4 release?
18:37 Whiteknight ENOVISION
18:37 allison "integration, interoperate, embed"
18:37 chromatic What are the milestones to work on with regard to that?
18:37 allison That is, fix the install problems.
18:37 Whiteknight hll interoperability got bumped from 1.3 I think
18:37 jhorwitz PDD10, for one.
18:38 allison Make it possible to load foreign HLL libraries.
18:38 allison And the extending/embedding interface.
18:38 Whiteknight what do we mean here by "integration"?
18:38 chromatic Okay, do we have a list of install problems and the approach on the wiki somewhere?
18:38 pmichaud install problems have been discussed on the trac tickets and mailing list, primarily.
18:39 pmichaud I don't know if there's a wiki page yet.  I have some suggestions but need a couple of hours to type them up.  I can briefly summarize here with "parrot's build process should create something that looks a lot more like the installation"
18:39 chromatic Would a wiki page help?  I know I could take on a couple of small pieces if I had some guidance.
18:40 pmichaud I'll be happy to contribute to a wiki page.
18:40 Whiteknight a wiki page would help
18:40 allison We have had several email messages on the subject.
18:40 pmichaud we had some good discussions about this at pvmw and later at yapc (more)
18:40 Whiteknight do we have anybody with knowhow and tuits available to focus on installation issues?
18:40 allison In one I suggested changing the directory structure of the repository to match the install.
18:40 pmichaud allison: at pvmw we suggested an intermediate step -- namely, have the result of "make" create something that looks like an install
18:41 pmichaud in particular, that would be the version that gets used for testing.
18:41 allison a, build/ directory, perhaps?
18:41 pmichaud could be that
18:41 chromatic Let's not get into technical solutions here; just project management and priority and resource allocation.
18:42 allison did anyone take notes at the pmvw/yapc discussions?
18:42 allison pvmw
18:42 pmichaud I have some notes -- I'm short of tuits for writing them up until later today
18:42 allison later today is good
18:42 allison or, dump the rough notes in a wiki for others to clean up
18:42 chromatic If they're on the wiki, can we collectively agree to work on those in the next week?
18:43 pmichaud knowing that there's general agreement on "refactor parrot build to look more like install" is a big help.
18:43 allison I'll help
18:43 smash (i can help too)
18:43 Util kid51 took detailed notes of the YAPC Parrot Implementors BOF.
18:43 pmichaud I have pictures from that BOF, too.
18:44 allison excellent
18:44 pmichaud (the BOF didn't discuss install directly, fwiw)
18:44 japhb Util: url?
18:45 Util japhb: I don't know if he has posted them yet.
18:45 japhb Util: ah
18:45 chromatic Is installation our top priority?
18:45 pmichaud It needs to be, yes.
18:45 allison yes
18:46 allison well, that needs clarification
18:46 allison fixing the problems that prevent building tcl/rakudo from parrot packages is a top priority
18:46 Whiteknight do we have a concise list of such problems?
18:46 smash which packages ?
18:47 allison that doesn't include every install feature/fix we might want to make
18:47 japhb And you also said that foreign HLL libraries was high on the list.  How high?  #2?
18:47 allison so we can focus in on the critical ones)
18:47 allison japhb: yes #2
18:47 pmichaud I will make a list of problems.  But #1 on the list is that it's currently very difficult to build dynpmcs and dynops
18:47 allison the embedding interface is #3
18:47 allison pmichaud: but it's possible
18:47 cotto smash, basically they need to be able to build from an installed parrot without any dependencies on the build tree
18:47 pmichaud and I have fundamental disagreement with the idea of "parrot should generate rakudo's makefile"
18:48 chromatic To summarize then.
18:48 allison pmichaud: impossible problems are a higher priority than "difficult"
18:48 chromatic pmichaud will create a wiki page listing the problems, focusing on the Rakudo/Tcl blockers.
18:48 pmichaud allison: at present I'm not able to build a testable version of rakudo from an installed parrot.
18:48 chromatic We've collectively agreed to address those problems as our top priority.
18:48 chromatic Objections?
18:48 pmichaud I've followed the tcl template about as far as I can go.
18:49 smash pmichaud: rakudo should be generating it's own makefile
18:49 allison pmichaud: Parrot doesn't build the makefiles, it just provides a utility to help. But there's no reason languages shouldn't write their own makefile processing.
18:49 pmichaud smash: I agree.  But the steps required to build those makefiles (especially the dynpmc and dynops portion) is particularly hairy.
18:49 pmichaud and requires a lot of maintenance and copy-paste from the language maintainer.
18:49 allison pmichaud: think of it like the language shell generators, use them when needed, and not when unnecessary.
18:50 pmichaud allison: I do think of it that way.  My points remain.
18:51 allison I have some comments here, basically in agreement with pmichaud, and with suggestions for how to move forward. In the new meeting format, should we have this discussion here or on #parrot?
18:51 chromatic #parrot
18:51 allison Okay, I'll hang around there and continue this after #parrotsketch.
18:51 pmichaud Same.
18:51 chromatic Second priority: loading foreign HLL libraries.
18:51 japhb Who is interested in working on the foreign HLL library problem besides Tene and I?
18:52 chromatic Do we have a page listing open issues and design notes?
18:52 japhb https://trac.parrot.org/parrot/wiki/HllInteroperability
18:52 japhb IS that what you meant?
18:52 japhb It's a little out of date (doesn't include the sub-namespace issue)
18:53 chromatic Who can update it?
18:53 japhb Assuming edit rights, Tene or I.
18:53 chromatic Okay, please do.
18:53 amuck_ joined #parrotsketch
18:54 Whiteknight japhb: what kind of help do you need?
18:54 Whiteknight sign me up as a willing participant
18:54 japhb Whiteknight: some of it is design ... some of it is testing to figure out what we haven't designed yet (the main purpose of OpenGL::Math).  And then there's some of just
18:55 japhb coding the design up for each HLL's part of the interface
18:55 japhb (In other words, making each HLL keep up its end of the bargain we're designing)
18:55 chromatic japhb, can you update the page sometime today?
18:55 japhb chromatic: yes
18:55 chromatic Excellent, thank you.
18:55 Whiteknight japhb++
18:55 chromatic Priority #3: extension/embedding API.
18:56 chromatic Do we have a page for this?
18:57 chromatic Do we have a jhorwitz?
18:57 jhorwitz here
18:57 chromatic Were you working on this?
18:57 jhorwitz sorry, have another call in a minute
18:57 jhorwitz yes
18:58 chromatic Okay, let's move on then.
18:58 chromatic pmichaud, you had a question.
18:58 pmichaud Whiteknight answered it earlier, but I have a new question.
18:58 pmichaud Or perhaps a comment.
18:58 pmichaud At any rate -- Parrot seems to have become much less stable for Rakudo of late.
18:59 pmichaud See, for example, http://nopaste.snit.ch/17066, which is today's daily spectest run.
18:59 chromatic Can we bisect that to a likely range of Parrot commits?
18:59 pmichaud I suspect this one is a pbc_to_exe issue.  masak++ noticed something similar as well.
19:00 pmichaud In *this* particular case, Rakudo was being run from the 1.3.0 copy of Parrot.
19:00 pmichaud It's entirely possible that a local Rakudo change triggered the issue (more)
19:01 pmichaud but notice particularly how many strings are being corrupted, such that "Int' is becoming 'Inu', 'Hash' is becoming 'Hasi', 'Num' becomes 'Nun', etc.
19:01 spinclad string++
19:01 Whiteknight that's weird and unpleasant
19:01 * Coke ~~, belatedly
19:01 pmichaud I noticed this same problem a week or so ago when I started adding "rw" properties to variable containers.
19:01 pmichaud i.e., simply attaching the property caused strings to be corrupted
19:02 allison so it could be a problem with properties?
19:02 pmichaud potentially.
19:02 chromatic That's likelier than GC.
19:02 pmichaud The problems do disappear when parrot is run with the -G flag, though.
19:02 allison properties are largely unused and only lightly tested
19:02 Whiteknight yeah, the GC is bulletproof
19:02 NotFound I remember a similar looking problems with strings months ago, don't remember well, maybe imcc related
19:03 Whiteknight so we need to add more tests for properties?
19:03 chromatic If you have a small test case in PIR, Whiteknight and I can tackle it.
19:03 pmichaud it's almost impossible to produce "small test cases in PIR" that tickle these bugs.
19:04 Whiteknight then get us a big PIR case and I'll narrow it down
19:04 Whiteknight never met the problem I couldn't solve with a little elbow grease and OCD
19:04 jevin joined #parrotsketch
19:05 pmichaud Whiteknight: will do.
19:05 Whiteknight thanks
19:05 pmichaud More generally I just wanted to note that somewhere since March we've begun losing our "stable platform for HLL implementors"
19:05 chromatic More tests needed!
19:06 pmichaud somewhere in contexts, strings, and imcc there's something that appears to be corrupting memory.
19:06 allison pmichaud: remember, that stability is the "supported" releases, not the monthlies
19:06 spinclad .o( 'rw' property makes ++typename ? )
19:06 pmichaud allison: if the problem isn't fixed in the next 3 weeks, it will be a problem in the supported release.
19:06 Coke allison: yes, but if there are bugs now, they're going to be in 1.4
19:06 allison so let's get it fixed now :)
19:07 Coke I'd also like to second pmichaud's point that it's not always possible to reproduce the bugs with a small amount of pir.
19:07 Coke (esp when we use dynops and dynpmcs)
19:07 pmichaud I guess what I'm saying is that resolving this issue feels like it ought to be a priority 1.5 or something like that.
19:07 moritz in the perl 6 test suite I noticed that real world apps and test cases test very different things, usually
19:07 chromatic I'm happy to make it a top priority.
19:08 moritz maybe somebody volunteers to write something bigger for the parrot test suite?
19:08 japhb moritz: QUITE.
19:08 Coke one of the "install" problems that tcl is facing is that pbc_to_c fails for the installed parrot. need that.
19:08 pmichaud I'm currently running rakudo against parrot trunk to see what errors crop up.
19:08 pmichaud (so far there are several)
19:08 moritz not geared towards one or two features, but using many, and make them interact in a typical way
19:08 allison Coke: yes, that's a big one
19:09 pmichaud anyway, that's all I had on that point.  I have another comment/question, though.
19:09 spinclad moritz: check in november, web.pm, etc as test suites?
19:09 allison Coke: afaik, that's still TT #495
19:09 Coke I don't think checking in those to parrot is a good idea, necessarily.
19:10 moritz spinclad: that might be a bit too big, and hard to keep up to date
19:10 Coke I think that having smoke tests of HLL projects feed back to parrot is a good idea, though.
19:10 chromatic Maybe we should add a milestone to improve our test coverage.
19:10 allison we don't necessarily need all tests checked into the repository, we could have an external "application test"
19:11 Coke pmichaud: one thing that would be helpful is bisecting to see if you can show a parrot revision introduced a bug (with a single revision of rakudo)
19:11 japhb moritz: perhaps some tests that push the boundaries of sanity?  Like pushing recursion limits (but not in theory breaking them), or creating and destroying millions of GC'able objects (but in theory only using a relatively small pool of "active" objects), and so on?
19:11 Coke I know that's not perfect, since your parrot requirement changes, but that's a good task for someone with tuits that doesn't require a lot of brainpower.
19:11 pmichaud Coke: that's what is really nasty/pernicious about the current situation.  Rakudo has been using the 1.3.0 release of Parrot since it came out.
19:11 moritz japhb: maybe. I have no really good ideas, just wanted to throw in one point of data/experience
19:12 pmichaud Coke: i.e., all Rakudo testing since 06-16 has been against the 1.3.0 release.
19:12 Coke and sometime since 06-16, you've started getting these errors?
19:12 pmichaud Coke: Yes.
19:12 pmichaud actually, that's not precise
19:12 pmichaud we were seeing some errors like these on 06-16, but the number and frequency of such reports is increasing
19:12 moritz I've seen some parrot backtraces in rakudo's spectest before parrot's 1.3.0 release
19:13 pmichaud correct.
19:13 pmichaud part of the problem with bisecting parrot before 1.3.0 is that there was a longish period of time where Parrot was effectively running with gc disabled
19:13 pmichaud so we don't really know where problems were introduced
19:13 Coke pmichaud: check against the last parrot before it was disabled and see if it's there.
19:14 pmichaud Coke: by "longish period of time" I mean six weeks.
19:14 Coke ok.
19:14 pmichaud There's no way that present-day rakudo will build against an eight-week-old Parrot.
19:14 Coke no, but you were getting failures with older rakudos, yes?
19:14 pmichaud and, of course, the Rakudo that existed eight-weeks-ago wasn't demonstrating the bugs.
19:14 chromatic Shall we try the PIR test case approach and see what happens and move on from there?
19:14 Coke ok.
19:14 Whiteknight pmichaud: what do you mean gc was disabled?
19:15 Coke Whiteknight: it wasn't running.
19:15 Coke chromatic flipped a comparison somewhere, I think.
19:15 spinclad bisect on parrots with gc reenabled...
19:15 chromatic r39215 was buggy.
19:15 japhb spinclad: you mean, bisect but always apply the gc reenable patch?
19:15 chromatic Let's move this discussion to #parrot if necessary to continue.
19:15 spinclad yes
19:16 chromatic pmichaud, you had another question.
19:16 pmichaud Perhaps this comes later in this meeting, but when/where do we attach priority and resources to the items in https://trac.parrot.org/parrot/report/16 ?
19:17 pmichaud so far they seem to be very starved.
19:17 pmichaud (also, somehow TT #389 was assigned to me, but I'm not the person who should have ownership of that ticket.)
19:17 Tene joined #parrotsketch
19:17 allison pmichaud: looks like one for me
19:17 allison will change
19:18 chromatic Any discussion about priorities there?
19:19 chromatic Other than "Please fix my language"?
19:20 pmichaud At least for "perl6", the issue is not "Please fix my language" but rather "Parrot has a demonstrable and repeatable bug that is causing perl6 to fail.  Please fix it."
19:20 chromatic Noted.
19:20 chromatic Any other questions?
19:21 pmichaud I'm guessing the answer to my previous question is "fix it yourself, or suffer starvation"?
19:22 allison I want to say "core developers commit to review that report regularly"
19:22 allison I will, at least
19:22 chromatic Me too.
19:22 Whiteknight ditto
19:22 japhb Question: Should there be any particular settings of the TT fields to indicate issues that affect the cross-HLL system?
19:23 NotFound Can TT #661 be closed? Last time something said it was pending perl6 testing in windows, if I remenber well.
19:23 japhb (Because I'm probably going to be converting some of our collective problem knowledge to tickets over the next couple days)
19:23 chromatic There may be a way to write a report which exposes those, japhb.
19:24 Coke (cross HLL) selecting a language should be a multi-select, not a single-select.
19:24 Tene Coke: what?
19:24 japhb chromatic: sure ... just wondering if we need to add a "cross-hll" entry to the language dropdown, perhaps in concert with Component: library?
19:25 chromatic Seems useful.
19:25 Tene ah, trac
19:25 Tene nm
19:25 NotFound If not, please someonw write on the ticket about what is blocking in perl6.
19:25 Util Coke: s/single/multi/ == Schema change; very non-trivial
19:25 spinclad language: cross-HLL?  (add a menu entry?)
19:25 Coke Util: trac sucks.
19:25 allison japhb: I'll add one
19:25 japhb allison: thanks.
19:25 chromatic Any other questions?
19:26 allison Coke: but better than the alternatives :)
19:26 Coke allison: respectfully disagree.
19:26 Coke which reminds me, we have about 300 RT tickets to still close out/transfer.
19:26 NotFound Coke: 298 last time I checked
19:27 Whiteknight I'm sure many of those are so vague and unfixable that they can just be closed
19:27 chromatic Let's call it a day then.
19:27 Coke ~~
19:27 Coke left #parrotsketch
19:28 chromatic When you update the Wiki page with tasks and analysis, please update the topic list in #parrot.
19:28 NotFound Whiteknight: agree
19:28 chromatic for our priorities anyway
19:29 Util left #parrotsketch
19:30 chromatic left #parrotsketch
19:40 NotFound left #parrotsketch
19:41 darbelo left #parrotsketch
19:52 PacoLinux joined #parrotsketch
20:13 jonathan left #parrotsketch
20:48 moritz left #parrotsketch
21:10 amuck_ joined #parrotsketch
21:11 Whiteknight joined #parrotsketch
21:39 amuck__ joined #parrotsketch
23:14 amuck joined #parrotsketch

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