Camelia, the Perl 6 bug

IRC log for #parrot, 2010-01-31

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:08 wknight-away windows--
00:09 whiteknight threads are not bad things, but they need to be used properly
00:09 whiteknight and with an understanding of the intrinsic costs and complexities
00:10 NotFound Yes, but looks like lot of people thinks that they need only to install the last buzzword and magically all programmers will be able to multithread anything efortless
00:10 dngor Mauve has more threads.
00:10 whiteknight NotFound: true. We had a programmer at work who wrote a high-performance messaging utility with dozens of threads and the thing was garbage
00:10 whiteknight but the kid was fresh out of school and didn't understand the costs of it
00:10 whiteknight we rewrote the utility with two threads, and it zips
00:12 whiteknight like any tool, if used correctly threads can be great things. Used incorrectly they can be a huge problem
00:13 dngor Great power, great responsibility, etc.
00:15 NotFound Great power and great responsability is hard to reach when the goal in most business is having cheap and easily replaceable programmers
00:19 NotFound kid51: I think I finally managed to install all CPAN modules needed, making smoke now
00:19 davidfetter if a tool requires 20 years' experience and an IQ of 200 to use properly, i have not problem calling that tool bad without qualification
00:19 davidfetter s/not/no/
00:21 kid51 NotFound:  Thanks.  We can always use more Win32 smokes.
00:22 NotFound davidfetter: What tool?
00:22 purl i guess tool is sufficiently extensible and flexible, so intelligent people produce tools that are supposed to help morons, meaning that the morons flock to the tool and abuse it
00:22 kid51 I think davidfetter is referring to threads.
00:22 davidfetter yep
00:23 * kid51 finished that article ... and understood some of it!
00:24 NotFound There are tools that some qualify as good for programming with threads. An Ada compiler, for example.
00:28 NotFound Failed to send the report, HTTP read timeout
00:29 kid51 NotFound:  Check the Smolder site.  You probably *did* send the report.  Smolder has been misbehaving like that for many weeks
00:31 kid51 You're there:  http://smolder.plusthree.com/ap​p/projects/report_details/32010
00:32 kid51 Here's your only test failure:  http://smolder.plusthree.com/ap​p/projects/tap_stream/32010/277
00:32 kid51 Do you know whether you are getting that error in trunk?
00:34 davidfetter NotFound, i confess i'll start caring about ada the first time i run across a project where it's involved :P
00:35 NotFound davidfetter: I don't care, just sing the song that people keeps playing since the '80 about those things.
00:37 NotFound I took a look at that paper. The tale about Moore's law and need for paralelism is always the same since then.
00:37 davidfetter parallelism, concurrency...these are real concerns. threads are not, imho, a good way to approach them
00:38 NotFound Funnily enough, the arguments about parallel programing with Cell were almost the same that at times of the Transputters.
00:39 NotFound I hoped that in 30 years thing changed a little more.
00:43 NotFound davidfetter: maybe, but I'd like to have more papers about good ways rather than re-exposing what's bad with threads ad nauseam.
00:44 davidfetter there are, as that paper mentioned, other ways to deal with the underlying issues. fixating on a known-broken technology strikes me as a terrible waste of time
00:44 NotFound If all progress in estructured programming had been repeating "goto is bad" we were still writing spaghetti code full time.
00:45 davidfetter hrm?
00:46 NotFound People worked on better ways, rather than wasting his time repeating how bas was goto.
00:46 NotFound s/bas/bad
00:46 davidfetter here i thought that was what parrot was for :)
00:46 davidfetter that == better ways
00:48 NotFound davidfetter: If some day we have at least that works fine even without any concurrency, that day I'll start worrying about paralelism ;)
00:48 NotFound at least a GC that
00:51 davidfetter i confess i'm a little biased. i do most of my work on a project that's strictly thread-free :)
00:54 NotFound BTW, parrot we can't implement a pseudo-threading based on switching contexts in the VM because of out old friends the inner runloops.
00:55 NotFound Well, we can, but will be almost useless.
00:55 davidfetter help me out here. what would make it useless?
00:56 NotFound If you need to wait for inner runloops to finish before switching contexts, most pseudothreads must wait forever.
00:57 NotFound If you don't wait... I failed to imagine how the thing can work.
01:02 davidfetter hrm
01:18 mikehh the problem I find with that paper is that it takes things to extremes
01:19 mikehh on the one hand he looks for deterministic processes (nothing is completely deterministic) and on the other
01:20 mikehh he considers the mapping for partial functions is not countable
01:21 mikehh that is like saying if you had a bunch of monkeys typing at random for infinite time one of them would produce the complete works of Shakespere or something like that
01:22 mikehh the problem being that we already have the complete works of Shakespere
01:22 davidfetter we do?!?
01:23 mikehh well maybe not - I am sure some were lost along the way
01:23 davidfetter here i thought there were a bunch of missing works
01:24 mikehh ah but they were not written by William Shakespere - just someone of the same name
01:24 mikehh a 16th Century identity crisis
01:29 mikehh I do agree however that threads (as currently implemented in Java or .Net) are not an ideal solution
01:32 NotFound Maybe we even still don't know what is the problem.
01:32 mikehh the concept of lightweight threads is better, but care is still needed - actually care is needed in any solution :-}
01:33 mikehh ah we have the answer but don't know the question
01:34 NotFound If the problem is 'give me a way to develop good and powerful programs with poorly trained programmers and low resources', maybe there is no solution until we have IAs HAL-900 class.
01:34 NotFound s/900/9000
01:35 LaVolta joined #parrot
01:35 NotFound And one can hope that such IAs will kill his managers }:)
01:53 dalek winxed: r381 | julian.notfound++ | trunk/Makefile:
01:53 dalek winxed: build setup.pir
01:53 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=381
02:15 treed What's the command to run a smolder test?
02:18 dukeleto make smoke
02:18 purl rumour has it make smoke is Smoke Testing: Run the test suite and send smoke.html to http://smoke.parrotcode.org/
02:18 dukeleto or make smolder_test
02:18 dukeleto treed: warning, the smolder server has been janky lately
02:19 treed janky like what?
02:19 * treed hasn't run a smolder in a while, figured he would
02:19 treed 32-bit OS X 10.5
02:20 dukeleto janky like it doesn't work
02:20 dukeleto the smolder server returns a 500
02:20 dukeleto the smolder server has memory leaks and other hidden bugs
02:20 treed nice
02:21 treed fulltest passes on my platform, for anyone interested
02:21 treed latest SVN
02:21 purl i think latest SVN is quite different, but i'll double check the dumper lines.
02:21 treed purl, forget latest SVN
02:21 purl treed: I forgot latest svn
02:21 treed latest SVN is the current new hotness
02:21 treed latest SVN?
02:21 purl latest SVN is the current new hotness
02:22 treed Files=341, Tests=12183, 150 wallclock secs ( 3.46 usr  2.04 sys + 72.12 cusr 40.08 csys = 117.70 CPU)
02:22 * treed gets dinner
02:25 dalek tracwiki: v6 | dukeleto++ | GitTransition
02:25 dalek tracwiki: http://trac.parrot.org/parrot/wiki/Gi​tTransition?version=6&action=diff
02:52 treed I don't suppose anyone's put in any work making it possible to subclass parrot;Class?
02:53 treed (How does Rakudo make its classes?)
02:53 treed Tene: ^
02:53 * treed is trying to pick up his cardinal thread, but the layers of indirection to do has-a instead of is-a is... headache-inducing.
02:54 cconstantine joined #parrot
02:59 treed $P1 = p6meta.'get_parrotclass'($P1)
02:59 treed Looks like it's also doing indirection.
03:01 treed Cardinal's branch iss41 has '!get_parrot_class' for part of the indirection
03:02 Tene treed: runtime/parrot/library/P6object.pir
03:02 treed I don't  have a runtime directory?
03:02 Tene treed: in the parrot repo
03:03 treed Oh.
03:03 Tene start on line 578
03:04 treed But, yeah, it looks like rakudo is also doing has-a
03:05 Tene What was your motivation for dropping P6object instead of subclassing it?
03:05 treed I don't want all the Perl6 methods.
03:05 treed Also Ruby isn't a prototype based language.
03:06 treed There's a whole crapload of things that it has that I don't want.
03:06 treed ALso there was a bug with it that never got responded to
03:06 treed So I gave up and worked around it.
03:06 treed http://trac.parrot.org/parrot/ticket/909
03:07 treed In general, any response I got when complaining about subclassing bugs was "Yeah, that's known to be broken. It'll get fixed eventualyl."
03:09 treed And I don't want to just work on other things, because the object model is kinda the underpinning of the entire language.
03:09 treed Or at least the stdlib.
03:09 treed I'd have to do fucked-up hackarounds to work on the stdlib, so I may as well get the foundation right and then I can build on it.
03:09 nopaste "kid51" at 70.85.31.226 pasted "kill_array_pmc branch: 2 failing test files" (164 lines) at http://nopaste.snit.ch/19422
03:10 treed ANd maybe eventually start using rubinius' libraries.
03:10 treed I can't remember if I bothered filing a bug about the fact that you can't throw subclasses of Exception.
03:11 treed Which is another pervasive feature that it'd be nice to get right in the beginning.
03:11 treed http://trac.parrot.org/parrot/ticket/835
03:12 treed also this, which was never responded to
03:12 treed also dealing with a subclass
03:22 patspam joined #parrot
03:32 kid51 msg Whiteknight see http://nopaste.snit.ch/19422 re two failed test files in kill_array_pmc branch
03:32 purl Message for whiteknight stored.
03:33 kid51 msg Whiteknight tried simply replacing 'Array' with 'ResizeablePMCArray', but that gave similar results
03:33 purl Message for whiteknight stored.
04:03 eternaleye joined #parrot
04:19 LaVolta joined #parrot
05:06 chromatic joined #parrot
05:32 treed I forgot what mailing list I was supposed to send that summary to.
05:32 treed I can only remember four items.
05:32 treed But, honestly, it gets pretty old doing the old RPG run-around trying to get shit working.
05:33 treed "Go to the castle and get the potion!"
05:33 treed "Oh, no, the way to the castle is blocked, you need to go find this old guy and get an axe."
05:33 treed "Oh, but he wants some shrubbery before he'll let you use his axe."
05:33 treed It's just like that.
05:33 treed At some point you start to lose all faith that you'll every get to the castle.
06:39 chromatic Milhouse: When are they going to get to the fireworks factory?
06:46 LaVolta joined #parrot
08:02 cotto_w0rk joined #parrot
08:27 fperrad joined #parrot
08:49 cotto joined #parrot
08:50 treed IS there a primer on using parrot's debugger anywhere?
08:50 treed I want to step through this code.
08:51 treed debugger?
08:51 purl i heard debugger was probably the better approach then since you can disect whos calling whom and how or perl's electronic teddy bear or 'perl -d' or a good way to see if anyone has done it already or http://dev.catalyst.perl.org/wiki/DebugSample or avalible with cperl-mode syntax highlight/editing goodness with emacs+gud+perldb or *really* unhappy to be inside init_declare
08:51 treed uh, hm.
09:02 treed Maybe -t 7 will be enough
09:02 treed A Visual Studio-style debugger would be really nice.
09:04 bacek joined #parrot
09:42 integral joined #parrot
09:50 jsut_ joined #parrot
10:01 bacek joined #parrot
10:04 bacek O hai
10:06 dalek parrot: r43659 | bacek++ | branches/kill_array_pmc/runti​me/parrot/library/Range.pir:
10:06 dalek parrot: Replace one more Array with RPA in Range
10:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43659/
10:06 dalek parrot: r43660 | bacek++ | branches/kill_array_pmc/runtime​/parrot/library/Test/More.pir:
10:06 dalek parrot: Wallparer problem in Test::More with comparision Null vs Undef. We probably need ticket for it to discuss proper solution of handling Nulls in Test::More.
10:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43660/
10:06 dalek parrot: r43661 | bacek++ | branches/kill_array_pmc/t (2 files):
10:06 dalek parrot: Replace usage of Array with RPA in test_more.t and from_parrot.t
10:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43661/
10:10 bacek msg mikehh I fixed failures you reported for whiteknight
10:10 purl Message for mikehh stored.
10:13 mikehh joined #parrot
10:16 bacek mikehh, aloha!
10:17 mikehh hi bacek
10:17 bacek mikehh, can you retest kill_array_pmc please?
10:17 mikehh bacek: ok will do
10:18 bacek mikehh, thanks
10:43 mikehh bacek: on failure in benchmark_tests - examples/benchmarks/array_access.pir
10:44 mikehh s/on/one/
10:44 bacek mikehh, thanks. Which platform?
10:45 mikehh bacek: Ubuntu 9.10 amd64 - g++ with --optimize
10:51 bacek mikehh, found it. Easy fix
10:53 mikehh i got it working - you need to work on two files t/benchmaks/benchmark.t and examples/benchmarks/array_access.pir
10:55 dalek parrot: r43662 | bacek++ | branches/kill_array_pmc/example​s/benchmarks/array_access.pir:
10:55 dalek parrot: Drop benchmarking of Array.
10:55 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43662/
11:05 bacek mikehh, fixed. Both of them
11:10 bacek msg whiteknight kill_array_pmc doesn't particularly help... Check r43665
11:10 purl Message for whiteknight stored.
11:11 dalek parrot: r43663 | bacek++ | branches/kill_array_pmc/t/benchmark/benchmarks.t:
11:11 dalek parrot: And more tests removed.
11:11 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43663/
11:11 dalek parrot: r43664 | bacek++ | branches/gc_encapsulate/src/pmc_freeze.c:
11:11 dalek parrot: Temporary fix to enable compilation again
11:11 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43664/
11:11 dalek parrot: r43665 | bacek++ | branches/gc_encapsulate (35 files):
11:11 dalek parrot: Merge branch kill_array_pmc into gc_encapsulate.
11:11 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43665/
11:17 cotto joined #parrot
11:34 mikehh bacek: all tests PASS - up to fulltest in kill_array_pmc branch at r43665 - Ubuntu 9.10 amd64 (gcc with --optimize)
11:34 bacek mikehh, excellent
11:38 mikehh interesting t/pmc/eval.t PASS - see TT #1142 - it doesn't have the GC mark/collect in the pbc code (t/pmc/eval_12.pbc)
11:38 mikehh that's in testr
11:39 bacek just slightly different GC pattern...
11:42 mikehh anyway need to get some dog food - bbl
12:05 ruoso joined #parrot
12:08 joeri joined #parrot
12:18 LaVolta joined #parrot
12:33 payload joined #parrot
12:50 cotto joined #parrot
13:08 payload joined #parrot
13:19 kid51 joined #parrot
13:29 Whiteknight joined #parrot
13:38 Whiteknight good morning #parrot
13:39 kid51 good morning whiteknight
13:40 Whiteknight kid51: where did that failure happen, make test?
13:40 kid51 Yes.
13:41 kid51 But subsequent commits by bacek++ appear to have corrected them.
13:41 kid51 Have a PASS on one box; awaiting another.
13:41 Whiteknight I'm updating and running again here
13:42 kid51 See bacek's commit messages as well.
13:43 Whiteknight yeah, I'm slowly digging through everything
13:44 kid51 kill_array_pmc: Linux/i386: HEAD PASS:  http://smolder.plusthree.com/app/pu​blic_projects/report_details/32016
13:44 Whiteknight nice
13:44 Whiteknight thanks for the tests
13:45 kid51 So AFAICT there's no reason inside Parrot not to merge
13:45 kid51 But HLLs may be a different story
13:46 Whiteknight I've tested Matrixy with no problems, so I don't think PCT is using Array
13:47 kid51 kill_array_pmc: darwin/ppc: HEAD PASS
13:59 Whiteknight purl msg bacek it did appear to help, now segfaulting in a different place.
13:59 purl Message for bacek stored.
14:12 Whiteknight I hate how visit_info pretends to have  VTABLE but isn't a PMC
14:19 dalek winxed: r382 | julian.notfound++ | trunk/winxedst (2 files):
14:19 dalek winxed: predef sprintf
14:19 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=382
14:35 cconstantine joined #parrot
14:36 szabgab joined #parrot
14:42 bit-man left #parrot
14:58 payload joined #parrot
14:59 dalek parrot: r43666 | whiteknight++ | branches/gc_encapsulate (3 files):
14:59 dalek parrot: copysize was not being set in gc_ms_reallocate_buffer, causing a segfault in memcpy. Fix that so now it is initialized. Fixes segfault. Miniparrot now builds config_lib.pasm properly, but the build fails several steps later with a built Parrot
14:59 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43666/
15:20 davidfetter joined #parrot
15:27 Whiteknight purl msg bacek config.fpmc is weird. It contains several k of NULL bytes and then some weird text that looks unrelaed to the config PMC. I think miniparrot is not freezing it correctly. Looks like the memory buffer might be getting recycled. Possibly a GC problem?
15:27 purl Message for bacek stored.
15:28 jan joined #parrot
15:33 jan_ joined #parrot
15:35 Psyche^ joined #parrot
15:42 dalek winxed: r383 | julian.notfound++ | trunk/winxedst0.cpp:
15:42 dalek winxed: avoid a few warnings in stage 0
15:42 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=383
15:56 dalek winxed: r384 | julian.notfound++ | trunk/config (3 files):
15:56 dalek winxed: testing configure
15:56 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=384
16:07 iblechbot joined #parrot
16:19 davidfetter is there some kind of frequently updated list that basically says, "parrot at least builds and passes tests here?"
16:22 Util davidfetter: PLATFORMS file in root dir
16:23 davidfetter hrm
16:23 davidfetter is there some kind of dashboard for smoke?
16:23 * davidfetter thinking of something like this: http://pgbuildfarm.org/
16:25 davidfetter btw, that buildfarm code is getting generalized at the moment, and will probably get released soon as a consequence :)
16:28 Coke cotto: pong
16:29 davidfetter buildfarm is full of "continuous integration" goodness
16:49 cygx joined #parrot
16:51 kid51 joined #parrot
16:52 cygx small question about building parrot with MinGW:
16:52 cygx it should work without MSYS, right?
16:53 theory joined #parrot
16:54 Coke cygx: I'm not sure; when building parrot on windows I typically use the ming that comes inside strawberry perl, and I'm not sure if msys is bundled in there also.
16:55 cygx Coke: according to README_win32, it should - but I get errors for missing header files which are nont part of MinGW
16:58 Coke k. I am sadly at work atm, so can't help debug this; if adding "requires MSYS" to the readme is an easy win, I can do that, though. =-)
16:58 Coke (if it turns out to be more complicated, a trac ticket is probably the way to go.)
16:58 cygx actually, I think the problem might be that I use the cygwin version of perl to run Configure.pl, so it pulls the wrong information
16:58 cygx so strawberry perl is the way to go?
16:59 Coke IME, yes.
17:00 cygx ok
17:00 Coke I tried cygwin for some time, and we had issues for the longest time. (Most since resolved, I thought.) I don't have access to the pay-for compilers. strawberry perl was an easy win, and it seems to JustWOrk.
17:01 * Coke is reminded to at least try to build parrot on his work win32 box.
17:02 cygx ok; I have a somewhat non-standard setup (standalone versions of both MinGW and Cygwin)
17:02 cygx a pure Cygwin build seems to work, but 'make test' reports failures...
17:12 dalek winxed: r385 | julian.notfound++ | trunk/winxedst1.winxed:
17:12 dalek winxed: conditional operator in stage 1
17:12 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=385
17:17 Coke cygx: I'm not sure we have any dedicated cygwin porters at the moment; if you can track down which ones are failing and open a ticket, that'd be a good start. (also, submitting smolder reports is helpful)
17:17 Coke (make smolder)
17:29 nopaste "cygx" at 84.176.96.243 pasted "parrot 2.0.0 cygwin test summary" (11 lines) at http://nopaste.snit.ch/19424
17:30 cygx hm...: make: *** No rule to make target `smolder'.  Stop.
17:33 kid51 cygx:  make smolder_test
17:33 kid51 Note:  Our Smolder site has been behaving somewhat badly for several weeks.
17:34 kid51 At the completion of your smoke test, it will probably successfully upload your report ...
17:34 kid51 ... but then it will report back to you that it got a timeout error.
17:34 kid51 However, within about 5 minutes your test should actually show up on smolder.
17:34 kid51 smolder?
17:34 purl smolder is http://sourceforge.net/projects/smolder or web-based smoke test aggregator used by developers and testers to upload (automated or manually) and view smoke/regression tests using the Test Anything Protocol (TAP). or http://smolder.plusthree.com/app​/public_projects/smoke_reports/8
17:35 kid51 cygx:  Could you paste output of:  prove -v t/configure/033-step.t
17:36 NotFound kid51: that means that my win32 attempt yesterday was succestuk?
17:36 NotFound succesful
17:36 kid51 Yes:  successful!
17:37 kid51 In fact, I pasted the link to your report on channel.
17:37 nopaste "cygx" at 84.176.96.243 pasted "prove -v t/configure/033-step.t" (63 lines) at http://nopaste.snit.ch/19425
17:37 cygx busy with fulltest - I'll send a smolder report after that...
17:38 kid51 NotFound:  I note that in that branch you had an error in one test.  I think that cygx just reported the same failure.
17:38 kid51 we should double-check.
17:39 NotFound kid51: yes, I noticed that but forgot to mention it.
17:39 kid51 cygx:  Those test failures are non-fatal
17:39 kid51 cygz:  But since I had a hand in writing t/configure/033-step.t, I will take a look at it.
17:44 Util davidfetter: Maybe the "TapTinder - Parrot build status" is what you want? http://tt.ro.vutbr.cz/ and http://tt.ro.vutbr.cz/build​status/pr-Parrot/rp-trunk/
17:45 kid51 NotFound:  Do you get that same test failure on Win32 in trunk?
17:45 davidfetter Util, interesting. i confess i'm more used to what the buildfarm does
17:45 kid51 (If so, then we know that the code in the branch is not the cause of the failure.)
17:49 kid51 cygx:  what do you get for:  perl -e "print qq|$^O\n|"
17:50 Coke (... why is that target called smolder_test and not smolder?)
17:50 Coke (seems like the _test is just extra verbosity)
17:54 cygx kid51: cygwin
17:58 kid51 Let me give you a patch to try out.
18:01 nopaste "kid51" at 71.246.120.241 pasted "t/configure/033-step.t: patch to handle Cygwin better" (40 lines) at http://nopaste.snit.ch/19426
18:02 kid51 Can you give that a spin?
18:02 NotFound kid51: don't know, will test when I can.
18:02 kid51 NotFound:  thanks -- I was actually directing that question to cygz, but I appreciate your testing efforts as well :-)
18:10 cygx kid51: fulltest is still running; I'll give it a try when its finished...
18:11 szabgab joined #parrot
18:13 kid51 Thanks.
18:13 Whiteknight Any rakudo people around?
18:13 Whiteknight Coke: ping
18:14 kid51 NotFound, cygx:  I just created TT #1422 re Parrot_IO.t failure on your boxes
18:21 cygx kid51: after patching, prove -v t/configure/033-step.t says: All tests successful.
18:21 kid51 Thanks.
18:22 kid51 My hunch was that on Cygwin, as on Win32, executables need to have the '.exe' suffix.
18:23 NotFound You can use .COM also, to remember old times ;)
18:24 cygx btw, the fulltest only reports the Parrot_IO failure...
18:24 Coke isn't file modified resolution time on windows coarser than normal? Are you testing something that is unreasonable to test?
18:24 Coke s/normal/on linux/
18:24 Coke Whiteknight: yes?
18:24 Whiteknight Coke: I was wondering if you would be able to test partcl with the kill_array_pmc branch
18:25 Coke Whiteknight: perhaps in several days after work burnout and recovery.
18:25 Coke @work right now.
18:25 Whiteknight gotcha. Maybe I'll try to do it myself before that
18:26 Coke i'll try it if I can. I probably want to setup something to test it on some arbitrary branche.
18:27 kid51 Coke:  re resolution time:  That sounds possible.  I observe that we get this failure on some Win32 Smolder reports but not others.
18:27 kid51 But I'm not actually very familiar with that test file, myself.
18:27 dalek TT #1422 created by jkeenan++: t/perl/Parrot_IO.t: 'not modified_since' test failing on some Win32 and ...
18:28 dalek winxed: r386 | julian.notfound++ | trunk/winxedst0.cpp:
18:28 dalek winxed: initial implementation of conditional operator in stage 0
18:28 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=386
18:28 dalek parrot: r43667 | jkeenan++ | trunk/t/configure/033-step.t:
18:28 dalek parrot: Expand provision for names of executables on Win32 to cover Cygwin as well.
18:28 dalek parrot: cygx++ for assistance in testing.
18:28 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43667/
18:30 Coke (win32) might be cygwin specific: http://www.cygwin.com/ml/cygwin​-patches/2002-q2/msg00308.html
18:31 Coke (that's just the closest thing I've been able to google.)
18:32 kid51 Coke:  but NotFound reported it last night on Win32
18:32 Coke could be a win32 thing. ISTR something about a 2-second resolution on such things, but could be FOS.
18:36 darbelo joined #parrot
18:36 * darbelo hates blackouts.
18:39 * davidfetter wonders whether darbelo is anywhere near EZE
18:39 darbelo EZE?
18:39 davidfetter airport code for buenos aires
18:39 purl buenos aires may be: SABE MRBA SABA SAEZ
18:40 darbelo Ah. No, not really near Ezeiza.
18:41 davidfetter at this distance, "near" would mean, "within an hour's non-flight transport of"
18:41 darbelo I'm actually 40-50 km away from the city proper.
18:41 darbelo It's about an hour and a half by train.
18:42 davidfetter whee!
18:42 * davidfetter has only been to .ar twice so far
18:42 darbelo And on the oposite direction from that airport. So add the time to cross the city.
18:43 cygx kid51, Coke: afaik the 2-second resolution is a FAT restriction, which would explain why it only occurs in some reports (I'm building from a FAT32 partition...)
18:44 kid51 I wonder if there would be any way to detect whether someone was on FAT or not.
18:44 * davidfetter wonders whether there's a PHAT partition
18:44 davidfetter yo
18:50 Whiteknight ...ok, partcl here fails all tests
18:50 Whiteknight so that's not good
18:51 Whiteknight partcl apparently relies on Array
18:55 patspam joined #parrot
19:00 darbelo Whiteknight: did you msg me about failing tests on pmc_freeze_with_pmcs?
19:01 dalek parrot: r43668 | darbelo++ | branches/pmc_freeze_with_pmcs/src/pmc/imageio.pmc:
19:01 dalek parrot: Fix copypasta errors.
19:01 dalek parrot: Misapropiate the set_pointer() VTABLE to poke at thaw_ptr.
19:01 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43668/
19:01 dalek winxed: r387 | julian.notfound++ | trunk/winxedst1.winxed:
19:01 dalek winxed: use conditional op in a few places in stage 1
19:01 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=387
19:01 Whiteknight darbelo: yes
19:02 Whiteknight bump PBC_COMPAT, and a few errors disappear
19:02 Whiteknight not all, of course
19:02 darbelo Good to know, but I'm braking the build horribly as we speak ;)
19:03 darbelo OTOH, other than adding a new PMC, I haven't commited anything that should caise failures yet.
19:05 Whiteknight partcl appears to fail all tests in trunk
19:07 nopaste "cygx" at 84.176.96.243 pasted "cygwin test summary from external NTFS hd" (11 lines) at http://nopaste.snit.ch/19427
19:07 cygx using an NTFS partition didn't make the error go awy...
19:07 cygx does anyone know if the 2 second resolution applies there as well?
19:07 Whiteknight on my box, anyway
19:23 * darbelo is starting to hate svn.
19:31 darbelo .
19:37 darbelo Damm. is parrot.org down?
19:38 Whiteknight apparently
19:44 payload joined #parrot
19:46 darbelo Firs the blackout, now the site's down. I'm never going to get this branch going...
19:46 * darbelo AAAAARRRRGGGGHHHHs
19:47 Whiteknight which branch, still pmc_freeze_with_pmcs?
19:48 darbelo Yeah.
19:51 darbelo I think I've the conversion pretty much done now.
19:52 Whiteknight nice
19:52 darbelo I have a few more segfaults to track down, but I'm confident that they're fixable.
19:52 Whiteknight I'm surprised the conversion wasn't more straight-forward
19:52 Whiteknight are you remembering to mark the ImageIO PMC?
19:53 darbelo Of course, no freeze == no build. So, the branch is hosed util I track that down.
19:53 darbelo I shouldn't have to. The stack walker ough to pick it up for me.
19:54 Whiteknight don't rely on the stack walker
19:54 Whiteknight it's shit-tacular
19:54 darbelo Still it's segfaulting in hash.c, not freeze code.
19:55 Whiteknight if an ImageIO PMC was getting recycled into a hash, and somebody was still calling the vtables, it could get messy
19:56 darbelo Doesn't look like that. It's freezing a hash that 'splodes.
19:59 darbelo And the whole "can't do shit with the sever down" is starting to make me want to stich to git.
19:59 darbelo s/stich/switch/
20:05 darbelo left #parrot
20:06 darbelo joined #parrot
20:12 zostay_ joined #parrot
20:13 wagle_ joined #parrot
20:18 cotto_working joined #parrot
20:18 payload joined #parrot
20:18 cotto joined #parrot
20:19 Whiteknight darbelo: send me a patch with your uncommitted changes and I will test too
20:19 Coke OSUOSL reported a problem with one of the drive farms, causing parrotvm and others to be mounted readonly.
20:20 Coke Whiteknight: partcl hasn't been tested in some time. partcl-nqp might have better luck.
20:21 estrabd_ joined #parrot
20:22 Coke cygx: ah... thank you. I could not remember the details on that one.
20:22 Coke detect fat32 is possible. It's probably easier to just put in a sleep for 2s before modifying the file.
20:23 eiro__ joined #parrot
20:23 Coke (yes, that slows down everyone else. You could just do it on windows) (just add a comment that checking the FS type is preferred, but more complex than we have time to fix right now.)
20:27 Whiteknight joined #parrot
20:27 cotto joined #parrot
20:28 TimToady joined #parrot
20:28 Khisanth joined #parrot
20:28 Ryan52 joined #parrot
20:28 rhr joined #parrot
20:28 TonyC joined #parrot
20:28 purl joined #parrot
20:28 workbench joined #parrot
20:28 hudnix joined #parrot
20:28 jjore joined #parrot
20:28 nopaste joined #parrot
20:28 cxreg joined #parrot
20:28 cxreg joined #parrot
20:28 darbelo Whiteknight: http://nopaste.snit.ch/19428
20:29 darbelo That add the last two vtables to the PMC. Adds the supporting functions needed for them to the .pmc file. And then modifies pmc_freeze.c to go through the PMC interface.
20:30 darbelo That will go in as three separate patches when svn returns.
20:31 darbelo In that state miniparrot segfaults when trying to freeze a hash in config_lib.pasm
20:31 * darbelo goes afk
20:31 darbelo be back in abit.
20:31 KatrinaTheLamia joined #parrot
20:31 slavorg joined #parrot
20:38 * Whiteknight is running rakudo spectest against kill_array_pmc first. Going well so far
20:40 lichtkind joined #parrot
20:43 Whiteknight the computer groans and strains
20:48 Whiteknight ....done
20:49 Whiteknight kill_array_pmc passes all the tests I've thrown at it
20:53 GeJ Good morning everyone
20:56 Whiteknight hello GeJ
21:06 GeJ Hi Whiteknight
21:07 GeJ Just to be sure, is svn down or is it my craptastic DSL line prevents me from co-ing this morning?
21:07 Whiteknight parrot.org is donw
21:07 Whiteknight down
21:07 Whiteknight svn is down
21:07 GeJ Oh, okay then. Thanks.
21:10 Whiteknight Coke: ping
21:22 dalek winxed: r388 | julian.notfound++ | trunk/winxedst0.cpp:
21:22 dalek winxed: refactor conditions in stage 0
21:22 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=388
21:24 Whiteknight pmichaud: ping
21:24 Whiteknight (I'm trying to thnk of who all is on the board and mght have the ability to un-crap the server)
21:29 Whiteknight particle1: ping
21:36 Whiteknight I wonder if I can just contact the osuosl.org people myself
21:42 Whiteknight ...I sent an email anyway
22:08 ilbot2 joined #parrot
22:08 Topic for #parrotis now Parrot 2.0.0 "Inevitable" released! | http://parrot.org | Priorities: deprecated core PMCs and VTABLE entries | Roadmap: http://icanhaz.com/parrotroadmap | Latest modified TT's: http://icanhaz.com/parrotbugs
22:09 Coke Whiteknight: pong.
22:09 Coke Whiteknight: did you miss my send?
22:10 Coke (RESEND) OSUOSL reported a problem with one of the drive farms, causing parrotvm and others to be mounted readonly.
22:10 Coke presumably this is causing the trouble.
22:10 * Coke quotes the email:
22:10 Coke "Around 19:08 UTC today our one of our Xen iSCSI nodes became
22:10 Coke unresponsive, causing affected VMs to remount their filesystems read-only.
22:10 Coke We are working to restore service and will provide further updates when
22:10 Coke they are available."
22:11 Coke parrotvm was on the list of
22:11 Coke Affected Virtual Machines
22:21 dalek winxed: r389 | julian.notfound++ | trunk/winxedst1.winxed:
22:21 dalek winxed: refactor conditions in stage 1
22:21 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=389
22:23 Whiteknight Coke: thanks for the info. Looks like there was a netsplit a while back and I was on the sort end of it. I didn't see your send
22:26 Whiteknight parrot.org is back up
22:27 Whiteknight Coke: as a bit of safety information, the website and repo are being backed up regularly?
22:29 Coke I have no idea, but assume so.
22:30 lichtkind Andy: ping
22:32 Coke This just in: "Everything should be back to normal as of 2:30PM PST."
22:35 dalek winxed: r390 | julian.notfound++ | trunk/winxedst1.winxed:
22:35 dalek winxed: escape param names to avoid conflict with pir/pasm reserved words in stage 1
22:35 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=390
22:43 payload joined #parrot
22:48 dukeleto 'ello
22:52 patspam joined #parrot
23:10 abqar joined #parrot
23:12 lichtkind dukeleto: hello
23:50 * Coke is done with big-bad-work project. whee!
23:51 Coke Whiteknight: you said partcl no longer works again trunk, yes?
23:55 Whiteknight tests failed on my box, yes
23:56 Whiteknight my first time running the tests in a long time, I may have done something wrong
23:57 Whiteknight failed every single test, and failed them relatively quickly
23:58 Coke svn info::git?
23:59 Whiteknight eh?

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

Parrot | source cross referenced