Camelia, the Perl 6 bug

IRC log for #parrotsketch, 2009-07-07

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

All times shown according to UTC.

Time Nick Message
00:35 davidfetter joined #parrotsketch
03:44 cotto joined #parrotsketch
09:58 masak joined #parrotsketch
13:19 Whiteknight joined #parrotsketch
14:17 amuck_ joined #parrotsketch
15:19 pmichaud joined #parrotsketch
16:02 NotFound joined #parrotsketch
16:19 darbelo joined #parrotsketch
16:20 moritz joined #parrotsketch
16:43 Coke_segf_TT_776 joined #parrotsketch
16:44 Coke left #parrotsketch
16:44 Coke joined #parrotsketch
16:45 Whiteknight # What I did last week:
16:45 Whiteknight * Fixed one GC heisenbug that had been bothering Rakudo. Help from bacek++
16:45 Whiteknight * Fixed a related bug in the gc-debug core involving PMCs on the stack and runloop recursion
16:45 Whiteknight * Looked into a second GC heisenbug but is not resolved yet. Can't reproduce it reliably yet.
16:45 Whiteknight * Added some documentation to the GC api.c on request from Infinoid++
16:45 Whiteknight # What I am doing this week:
16:45 Whiteknight * Idle debugging, testing, closing tickets
16:45 Whiteknight * Hoping to close the io_cleanups branch, finally.
16:45 Whiteknight * Busy with $real_life this week, limited hacking time available.
16:45 Whiteknight # What I am blocking on:
16:45 Whiteknight * Nothing.
16:47 Coke Done:
16:47 Coke - partcl
16:47 Coke - Configure now checks parrot revision/release
16:47 Coke - rakudo
16:47 Coke - began sifting through RT.
16:47 Coke - trimmed down a rakudo segfault to plain PIR.
16:47 Coke Blocking:
16:47 Coke - partcl
16:47 Coke - TT #691 - can't build executable
16:47 Coke - TT #776 - segfault in <$I0 = isa $P0, 'Undef'>
16:47 Coke - https://trac.parrot.org/parrot/report/16
16:47 Coke - http://code.google.com/p/partcl/wiki/ParrotIssues
16:47 Coke .
16:49 Util joined #parrotsketch
16:58 NotFound * Use HLL mappings in Parrot_str_split.
16:58 NotFound * Working on exceptions handling, TT #804 / RT #63592.
16:58 NotFound * Usual minor clean and fix.
16:58 NotFound EOR
17:03 Whiteknight_ joined #parrotsketch
17:33 * darbelo is trying out the "link to a report" idea from the wiki.
17:33 darbelo Report at: http://www.parrot.org/content/not-dead-yet
17:33 darbelo EOR
17:35 * moritz has no report today
17:54 Util # Done:
17:54 Util * Investigated pmichaud++ issue of fakecutable failing when perl6.pbc succeeds; not reproducible on my hardware.
17:54 Util * Worked privately on tinkertoy PPI-based 5->6 translator; no public repo yet.
17:54 Util * Ran out of tuits early.
17:54 Util # Plan for the week (or so):
17:54 Util * Carrying-over all items from last week.
17:54 Util * Commit /etc/magic file for Parrot bytecode, from PWMW last month.
17:54 Util * Currently reading: "Modern dynamic linking infrastructure", and linked-to articles.
17:55 Util - http://lambda-the-ultimate.org/node/3474
17:55 Util # Blockers:
17:55 Util * Minimal tuits this week.
17:55 Util .end
18:03 allison joined #parrotsketch
18:04 cotto # What I did:
18:04 cotto * Figured out the old ops2c code enough to decide what opsc's file structure should look like.
18:04 cotto * Created a new ops_pct branch.
18:04 cotto * Committed some minimal scaffolding code (makefile template, opsc.pir and a bunch of empty files) to get make working.
18:04 cotto * Got a minimal parser working.
18:04 cotto * Sneakily got bacek++ to do a bunch of work extending the parser, adding tests and actions.
18:04 cotto # What I hope to do and how many tuits I expect to have:
18:04 cotto * I'll continue working on opsc.  It's reached the point where it's fun to work on, so progress should be good.
18:04 cotto * At a minimum, I expect to get the Op class written and integrated with the compiler.
18:04 cotto * The tuit outlook is good.
18:04 cotto # What could block my progress:
18:04 cotto * No blockers.
18:04 cotto .eor
18:12 mikehh joined #parrotsketch
18:15 fperrad joined #parrotsketch
18:16 fperrad * play with MinGW gcc 4.4.0
18:16 fperrad * propose a new Configure step for thread : see TT #815
18:16 fperrad EOR
18:21 pmichaud What I did:
18:21 pmichaud * $otherjob
18:21 pmichaud * Rakudo now passing 11,580+ spectests
18:21 pmichaud * As of yesterday, Rakudo failing 112 tests (mostly related to -G bugs)
18:21 pmichaud * Closed lots of Rakudo RT tickets for Rakudo day
18:21 pmichaud * Added some more operators to Rakudo setting
18:21 pmichaud * Fixed some PGE and PCT bugs
18:21 pmichaud * Lots of testing to support -G debugging
18:21 pmichaud * Had long and useful conversation with allison++ regarding building from installed Parrot and 'make install' issues
18:21 pmichaud * Lots of long-term planning and brainstorming
18:21 pmichaud What I'm doing this week:
18:21 pmichaud * 'make install' for Rakudo and Parrot
18:21 pmichaud * Carping about -G bugs
18:21 pmichaud * Rakudo day:  'rw' attribute on containers, BEGIN blocks (spill over from last week)
18:21 pmichaud * Planning events and presentations for OSCON, YAPC::EU
18:21 pmichaud * Completion of hll interop tasks for 1.4 release (this week and next)
18:21 pmichaud What I'm blocking on:
18:21 pmichaud * Parrot -G bugs / memory corruption -- major blocker
18:21 pmichaud * Still cannot build from installed parrot
18:21 pmichaud * Various perl6-tagged tickets in Trac
18:21 pmichaud Queue two items
18:21 pmichaud EOR
18:22 barney joined #parrotsketch
18:24 chromatic joined #parrotsketch
18:24 chromatic No progress on the profiling runloop (sorry, cotto).
18:24 chromatic Chasing down memory errors Rakudo triggers; some progress there.
18:24 chromatic Will continue to work on those.  Will attempt to get back on the profiling runloop.
18:25 allison - Ticket queue review, catching up on progress, answering questions.
18:25 allison - Testing Tcl and Rakudo from installed Parrot for build and install issues (looks like making the build paths match the install paths is only a small help, will also need configuration changes).
18:25 allison - Production process handholding on the book.
18:25 allison EOR
18:25 barney Pipp: Fixed a couple of OO-features
18:25 barney Tuits: Few, as I started new freelance project with OTRS AG
18:25 barney .eor
18:32 cotto hi
18:32 Coke ~
18:32 * Util is here
18:32 NotFound hola
18:32 pmichaud Hello.
18:32 fperrad Hello
18:33 darbelo Hola
18:33 barney hi
18:34 chromatic hello
18:34 allison hi
18:34 chromatic Did everyone report?
18:34 pmichaud reported.
18:34 chromatic Great.
18:35 chromatic Let's review our critical tasks.
18:35 chromatic #1 installation -- any progress?
18:35 pmichaud Allison and I had a very useful discussion on the topic last week after #ps
18:35 pmichaud I think we're in agreement that ideally Parrot's "make" will result in something that looks much closer to installed parrot
18:36 pmichaud One of my questions is whether that's to happen before 1.4, or after.
18:36 pmichaud I strongly prefer before, and I'm willing to dedicate significant tuits to that.
18:37 chromatic Any objections to before?
18:37 allison before is fine, as long as it's soon
18:37 pmichaud I'm thinking to prototype it in a branch.
18:37 allison as in, before next parrotsketch
18:37 pmichaud If the branch is ready by next week, great.  If not, we say "we can't do it" and wait until after 1.4
18:37 allison sounds reasonable
18:37 chromatic What do you need to work on it?
18:38 pmichaud mostly just assurance that my tuits wouldn't be better spent elsewhere :-)
18:38 pmichaud it sounds like it's okay to go forward with it
18:38 pmichaud the biggest thing we'll need is platform testers
18:38 chromatic Are there parts where other people can assist?
18:38 pmichaud sure
18:38 pmichaud I'll outline what needs to be done in an email after parrot sketch
18:39 chromatic Great, thank you.
18:39 pmichaud then I'll start the branch
18:39 chromatic Shall we move to #2?
18:39 pmichaud yes.
18:39 chromatic #2 -- using foreign language libraries
18:40 pmichaud that's also on my list to complete this week
18:40 chromatic Who else is available to work on this?
18:40 pmichaud I don't see any problems with that.  Tene and others have been working on it already; I think it's just codifying some of what they've explored and chosen.
18:41 chromatic That sounds hopeful.
18:41 chromatic I like hopeful.
18:41 pmichaud yes -- there's been steady progress, we just need to wrap things up
18:41 pmichaud (more)
18:41 pmichaud on the other hand, this is an area that is currently receiving significant attention on the Perl 6 side of things, so it's probably worth holding a day or two to see if those changes land
18:42 chromatic Is there a list of open tasks?
18:42 cotto pmichaud, you mean holding back the release if needed?
18:42 pmichaud No, not holding back the release.
18:42 cotto ok
18:42 pmichaud Holding off on wrapping the task up.
18:42 pmichaud Regardless it will be ready in time for release.
18:43 pmichaud I don't have a list of open tasks at the moment, no.
18:43 pmichaud It doesn't feel "significant", though.
18:43 chromatic If it starts to look significant by Saturday, can you make a list of tasks for the list?
18:44 pmichaud sure.
18:44 chromatic #3 -- external API
18:45 chromatic I don't see jhorwitz here; is anyone else working on this?
18:46 chromatic Okay, I'll ask him for a status update.
18:46 chromatic What other tasks are priorities?
18:46 chromatic I know the memory bugs Rakudo and Tcl are running into are.
18:46 chromatic Whiteknight and I will work on those.
18:47 pmichaud Those are big-time blockers for Rakudo right now.
18:47 NotFound Isn't was supposed to be a wiki page about the extern/embed thing?
18:47 Whiteknight just tell me what you need
18:47 pmichaud If they aren't fixed, I don't know what to do for Rakudo's July release.
18:47 pmichaud Short of releasing something that has known failures.
18:47 Whiteknight we'll get them fixed
18:48 pmichaud Given that I've been carping about these same issues for a couple of years, I hope you'll pardon me if I'm not hopeful at this stage.
18:48 pmichaud That said, I'm very encouraged and appreciative of the efforts of this past week.
18:48 pmichaud I've seen some truly impressive work, so I guess there probably is hope.  :-)
18:48 kurahaupo joined #parrotsketch
18:49 chromatic Other priorities?
18:49 pmichaud Are we likely to have the installed pbc_to_exe issues resolved before release?
18:49 chromatic I certainly hope so.
18:49 pmichaud Who is working on that?
18:50 chromatic I see some comments from Gerd.
18:50 chromatic Maybe Util and I can poke at it.
18:50 pmichaud it might "fall out" of the work I'm planning to do on an install refactor, but it might not.
18:50 Util I touched the ticket last. I can at least move it further along.
18:51 chromatic Should we wait to see what you're working on for installation?
18:51 pmichaud Given that there's some question as to whether installation will land by next week, I think it needs to go in parallel.
18:51 pmichaud because we really do need pbc_to_exe working, even if 'make' doesn't get refactored.
18:51 Util I am willing to work in parallel, and risk work being invalidated.
18:51 pmichaud I don't think it'll be invalidated.
18:52 pmichaud I think it'll end up being the same for the branch.
18:52 chromatic Alright, we'll work on it.
18:52 Coke I need to add a blocker.
18:53 chromatic Go ahead, Coke.
18:54 Coke apparently partcl would get more parrot developers hacking on related bugs but for the inability to build (easily) against both an installed parrot and a build-dir parrot.
18:54 Coke so, without that, my blockers have blockers.
18:54 Coke (thankfully, that issue seems to be on the list.)
18:54 chromatic Agreed.
18:54 pmichaud Coke:  making it easier to build easily against a build-dir parrot is the focus of my "make refactor"
18:57 chromatic Other blockers?
18:57 pmichaud I have one
18:57 pmichaud at least, a minor issue
18:57 chromatic go ahead
18:58 pmichaud currently when building/testing Rakudo against an installed parrot, an installed copy of dynops/dynpmcs will always take precedence over the locally built one
18:58 pmichaud is that by design or is that a bug?
18:59 Coke I vote bug.
18:59 allison it's a question of search paths
18:59 allison one or the other has to have priority
19:00 spinclad or specify $(WHICH_DIR)/dynops in makefile ?
19:00 pmichaud for library paths, the versiondir path is the last one searched
19:00 pmichaud for language paths, the versiondir path is the last one searched
19:00 allison pmichaud: it is searching the local dir last
19:01 pmichaud allison: right.  This is different from library/language
19:01 pmichaud so I'm wondering if that's by design or if it's just a bug
19:01 allison call it a bug in design
19:01 pmichaud if by design, it makes it harder to build a new version of rakudo if there's already a version installed
19:02 pmichaud because you always get the old dynext stuff
19:02 allison versiondir should still be last
19:02 allison and it should still follow the "runtime" dir search
19:02 pmichaud okay.  I'll fix versiondir.
19:02 allison (which is probably going away in your branch)
19:02 pmichaud yes, runtime should be going away in the branch.
19:02 allison but the one that's searching plain "dynext/" should be first
19:02 pmichaud in the sense of "not in the library search path"
19:03 pmichaud okay.
19:03 pmichaud If nobody beats me to them I'll fix.
19:03 chromatic Other blockers?
19:05 chromatic Alright, let's call this a meeting then.
19:05 chromatic Thank you everyone.
19:07 fperrad left #parrotsketch
19:09 amuck_ left #parrotsketch
19:11 Util left #parrotsketch
19:15 PacoLinux left #parrotsketch
19:16 darbelo left #parrotsketch
19:23 pmichaud left #parrotsketch
19:28 Whiteknight_ joined #parrotsketch
20:00 Coke left #parrotsketch
20:18 NotFound left #parrotsketch
21:16 Whiteknight joined #parrotsketch
22:13 Whiteknight joined #parrotsketch
22:29 mikehh_ joined #parrotsketch
22:38 mikehh_ joined #parrotsketch
23:40 particle joined #parrotsketch

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