Camelia, the Perl 6 bug

IRC log for #parrot, 2009-12-30

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:16 kid51 joined #parrot
00:28 Coke will anyone cry if I eliminate all the _DIR variables in the makefile?
00:28 Coke (or at least most)
00:28 Coke they seem to just be obfuscating.
00:28 kid51 I shed my last tears in 1993.
00:28 Coke excellent!
00:28 purl EGG-see-lent!
00:28 payload joined #parrot
00:29 kid51 But maybe you could paste a patch so we have a better idea of what you're plotting?
00:29 Coke it's all in one_make.
00:29 kid51 that's a branch?
00:29 Coke ayup
00:29 Coke trying to do general makefile cleanup. notes in #ps log today.
00:29 kid51 I took a too quick glimpse at that.
00:30 * Coke works on killing compilers.dummy first.
00:30 kid51 Was looking in from $job when all hell broke loose.
00:31 kid51 While I'm doing a checkout ... any thoughts on TT #1393?
00:34 Coke no, I saw that one. just standard diagnostics, does it still die if you disable GC, can you reduce the code needed to generate the signal, can you capture a backtrace in gdb...
00:34 kid51 I'll see what I can do.  Don't have much experience with C debugging.
00:35 kid51 Another ticket I'm puzzled about is http://trac.parrot.org/parrot/ticket/473 (see most recent comments)
00:37 kid51 Am I correct in believing that any and all impact of changes in the one_make branch should be evident by completion of Configure.pl?
00:43 Zak joined #parrot
00:43 Coke no
00:43 kid51 "does it still die if you disable GC?"  No, it does not!  So it's the --gc-debug that sparks the problems.
00:43 Coke I'm changing the Makefile, so you have to actually build parrot.
00:44 nopaste "kid51" at 70.85.31.226 pasted "TT #1393: Without --gc-debug, pge_examples.t PASSes under harness" (4 lines) at http://nopaste.snit.ch/19186
01:02 nopaste "kid51" at 71.255.54.187 pasted "gdb parrot on t/compilers/pge/pge_examples_2.pir" (18 lines) at http://nopaste.snit.ch/19187
01:07 kid51 Coke FWIW one_make branch passes all tests in 'make test' ... except the one that's giving me problems in trunk.
01:07 theory joined #parrot
01:10 Coke heh.
01:21 kid51 Never fails:  No sooner do I close a ticket for Smolder upload timeouts than I start getting hit with them myself!
01:23 kid51 Hmm, this is weird.  My 3 smoke tests actually *did* upload to Smolder ... but the feedback I got on the command-line was:
01:23 kid51 Could not upload report to Smolder at http://smolder.plusthree.com
01:23 kid51 HTTP CODE: 500 (read timeout)
01:23 kid51 make: *** [smolder_test] Error 255
01:23 kid51 in all 3 instances!
01:26 kid51 Coke:  My earlier report that there were no (significant) problems with one_make branch was probably wrong.
01:26 Coke hokay
01:26 Coke at this point, if it's passing any tests, it's ok. =-)
01:26 Coke but if you have specific things that are failing, I'm happy to hear them.
01:27 kid51 It built, but something's amiss during 'test'
01:27 kid51 Doing 'make clean' and new smolder
01:27 Coke probably missing some deps.
01:27 kid51 Yes.
01:30 nopaste "kid51" at 71.255.54.187 pasted "one_make branch: preliminary report on 'make test' failure" (7 lines) at http://nopaste.snit.ch/19188
01:31 Coke ah, yah, data_json is still part of compilers.dummy
01:31 Coke one of the last holdouts.
01:37 Andy joined #parrot
01:37 nopaste "kid51" at 71.255.54.187 pasted "one_make branch: preliminary report on 'make' failure" (9 lines) at http://nopaste.snit.ch/19189
01:38 Coke hurm.
01:41 Coke Doesn't look like compiler nqp's "boot" target is part of the normal build.
02:31 JimmyZ joined #parrot
03:12 bacek joined #parrot
03:24 Coke any reason we cannot rely on prove?
03:38 JimmyZ when it's #!parrot ?
03:41 JimmyZ Coke, I don't know whether it's GFWoC issue or not.
03:43 Coke GFWoC?
03:47 Coke anyone know how to make nqp_test work from the top level dir?
03:47 Coke (right now it does a cd into compilers/nqp)
03:48 JimmyZ GFW
03:51 Coke aha. Parrot::Test::Harness is smarter than I thought. fixed.
04:15 dalek parrot: r43313 | coke++ | branches/one_make (12 files):
04:15 dalek parrot: Convert compiler/nqp's recursive make to an include.
04:15 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43313/
04:31 dalek parrot: r43314 | plobsing++ | branches/pmc_freeze_cleanup/src/pmc (14 files):
04:31 dalek parrot: remove checking for EXTRA_IS_NULL in PMCs (it will always be so)
04:31 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43314/
04:47 dalek parrot: r43315 | coke++ | branches/one_make (8 files):
04:47 dalek parrot: convert data_json's recursive make to an include.
04:47 purl I don't know how to convert data_json's recursive make to an include..
04:47 dalek parrot: Last element of the dummy compiler target.
04:47 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43315/
04:48 Coke holy crap.
04:48 purl only in the Vatican, my friend
04:49 Coke The shenanigans required to fill in @library_tests@ instead of just leaving that in the Makefile is insane.
04:49 * Coke rips it out.
04:52 * Coke ughs.
04:52 * Coke reverts.
05:04 dalek parrot: r43316 | coke++ | branches/one_make (2 files):
05:04 dalek parrot: $(IMCC_DIR) -> compilers/imcc
05:04 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43316/
05:04 dalek parrot: r43317 | coke++ | branches/one_make/config/gen/makefiles/root.in:
05:04 dalek parrot: eliminate PF_DIR variable
05:04 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43317/
06:25 msmatsko joined #parrot
06:37 bacek joined #parrot
07:19 nopaste "tene" at 76.27.121.193 pasted "glxinfo" (633 lines) at http://nopaste.snit.ch/19190
07:19 Tene Ack.
07:19 Tene Didn't mean for that to go here. :)
07:19 Tene Too used to nopasting into this channel.
07:29 hejki :D
07:29 hejki but hey, at least your direc rendering works!
07:29 hejki s/rec$/rect/;
07:40 Tene Heh.
07:40 Tene I'm trying to play with http://love2d.org/ but it's segfaulting after X failures on my laptop.
07:40 Tene But works fine when forwarded to a different X server, like my desktop, or Xephyr.
07:40 hejki ahh.. pygame eqv for lua?
07:41 hejki looks nice. too bad i'm not that much into lua.
07:41 hejki ever played with bullet? :)
07:41 Tene Not sure.  Don't know much about pygame.  The love package actually has a separate binary, doesn't install any lua libraries, has some stuff for reading dirs out of a zip file, etc.
07:41 Tene No, never heard of it.
07:42 hejki ahh ok
07:42 hejki so it's different from pygame :P
07:42 hejki bullet is nice
07:42 hejki physics :)
07:44 hejki http://bulletphysics.org/
07:57 Essobi joined #parrot
08:05 eternaleye joined #parrot
08:14 mikehh joined #parrot
08:21 eiro_ joined #parrot
08:21 iblechbot joined #parrot
08:44 bacek joined #parrot
08:49 xenoterracide left #parrot
09:17 mikehh joined #parrot
09:46 mikehh joined #parrot
09:53 fperrad joined #parrot
09:54 mikehh joined #parrot
10:10 cognominal joined #parrot
10:39 mikehh joined #parrot
10:56 bacek joined #parrot
11:16 payload joined #parrot
12:07 mikehh joined #parrot
12:21 dalek parrot: r43318 | bacek++ | branches/boehm_gc (3 files):
12:21 dalek parrot: Get rid of GC_DEBUG macro. It's not so usefull and conflicting with
12:21 dalek parrot: Boehm GC.
12:21 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43318/
12:21 dalek parrot: r43319 | bacek++ | branches/boehm_gc/src/gc/gc_boehm.c:
12:21 dalek parrot: Implement finalization and explicit collect in Boehm GC.
12:21 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43319/
12:25 bacek joined #parrot
12:31 masak joined #parrot
12:36 ruoso joined #parrot
12:51 cognominal joined #parrot
13:11 fperrad joined #parrot
13:24 plobsing joined #parrot
13:52 whiteknight joined #parrot
14:06 bluescreen joined #parrot
14:37 payload joined #parrot
15:03 PacoLinux joined #parrot
15:06 pdcawley joined #parrot
15:10 bluescreen joined #parrot
15:14 patspam joined #parrot
15:28 mikehh joined #parrot
15:30 bluescreen joined #parrot
15:34 fperrad_ joined #parrot
15:48 ash_ joined #parrot
15:55 Psyche^ joined #parrot
16:06 payload joined #parrot
16:08 dalek parrot: r43320 | mikehh++ | branches/one_make/compilers (2 files):
16:08 dalek parrot: set svn properties
16:08 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43320/
16:08 dalek parrot: r43321 | mikehh++ | branches/boehm_gc/MANIFEST:
16:08 dalek parrot: regenerate MANIFEST
16:08 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43321/
16:14 plobsing question about constant PMCs - must they only contain constant strings as well as PMCs?
16:20 mikehh plobsing: no idea really, but it sounds right
16:21 mikehh I would think that if the PMC is constant it should not change
16:21 plobsing thats not what "constant" means in parrot.
16:21 plobsing "constant" means "ignored by GC"
16:21 plobsing roughly
16:22 mikehh then it should be called something else maybe
16:22 plobsing preaching to the choir
16:22 mikehh to me constant means immutable
16:25 dalek parrot: r43322 | mikehh++ | branches/boehm_gc (2 files):
16:25 dalek parrot: set svn keywords
16:25 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43322/
16:25 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#31381), fulltest) at r43319 - Ubuntu 9.10 amd64 (g++ with --optimize)
16:41 dalek parrot: r43323 | mikehh++ | branches/boehm_gc/src/gc/gc_boehm.c:
16:41 dalek parrot: fix codetest failure - at least one space between C keyword and subsequent open parenthesis
16:41 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43323/
16:41 dalek parrot: r43324 | mikehh++ | branches/boehm_gc/src/gc/gc_boehm.c:
16:41 dalek parrot: fix codetest failure - no c++ comments (highlighted with /*** .. */)
16:41 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43324/
16:41 lucian joined #parrot
17:06 payload1 joined #parrot
17:13 cotto mikehh, istr that bacek leaves those c++ comments there on purpose.
17:13 dalek parrot: r43325 | mikehh++ | branches/boehm_gc/src (3 files):
17:13 dalek parrot: fix codetest failure - Correctly indented preprocessor directives
17:13 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43325/
17:23 theory joined #parrot
17:47 fperrad_ joined #parrot
17:49 ruoso joined #parrot
19:27 Coke mikehh++ # fixing distro bugs.
19:28 Coke msg particle you're the only person I know who uses nmake. Can you try out the one_make branch?
19:28 purl Message for particle stored.
19:28 Coke seen pmichaud ?
19:28 purl pmichaud was last seen on #parrot 8 days, 3 hours, 18 minutes and 33 seconds ago, saying: good morning.  [Dec 22 16:07:55 2009]
19:29 ash_ Coke: do you use/have os x?
19:31 joeri joined #parrot
19:36 payload joined #parrot
19:38 Coke ash_: not until I resurrect my imac which just killed the boot disk.
19:39 Coke (but I'm stuck at 10.4)
19:39 ash_ joined #parrot
19:39 ash_ bah colloquy just crashed
19:39 ash_ Coke: okay, just curious
19:44 ash_ i noticed in the opengl stuff that parrot searches for OpenGL in the runtime/parrot/library/OpenGL.pir file in OS X it looks for a libGL.so.1, on my computer (10.6.2) it can't open the libGL librarys (which is actually named libGL.dylib, not .so.1)  but my i am misunderstanding how NCI works
19:48 bacek joined #parrot
19:57 japhb ash_, OpenGL.pir tries a series of possible library names until it finds one it can load.  The second one it tries is libGL, and on OS X, that should resolve to libGL.dylib.  If it doesn't (anymore, perhaps), then we just need to add 'libGL.dylib' to the list at line 114.
19:57 japhb I have to say, OS X linking is a complete nightmare.
19:57 dalek parrot: r43326 | mikehh++ | branches/boehm_gc/src/gc/alloc_memory.c:
19:57 dalek parrot: fix codetest failure - there should be one space or a newline after a comma and unwrapped macro arguments
19:57 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43326/
20:09 ash_ japhb: hm, here's the error i get
20:09 ash_ Could not find a suitable GL shared library! which is thrown on that section of code (around 114)
20:11 japhb Yeah, that means it couldn't find a match.  Try copying line 114 and adding the .dylib on one copy, and 'make reconfig'
20:14 dalek parrot: r43327 | mikehh++ | branches/boehm_gc/src/gc/gc_boehm.c:
20:14 dalek parrot: fix codetest failure - add missing function documentation (this needs clarificartion - placeholder)
20:14 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43327/
20:15 ash_ do i need to run make after reconfig?
20:15 japhb ash_, yes, sorry, forgot to mention that
20:15 ash_ kk, i figured as much, its running right now
20:15 * japhb is multitasking heavily right now
20:17 Coke mikehh: you can add a function to the skip list for that instead of adding a potentially incorrect description.
20:20 ash_ japhb: still it doesn't work
20:20 purl It's a Y2K error!  Panic!  Sue!
20:20 japhb :-(
20:21 japhb I remember something goofy about having to link parrot itself with the frameworks for OpenGL and GLUT ... we had that working at one point; I wonder if that got broken
20:21 ash_ oh, its looking in the wrong folder
20:21 japhb ?
20:22 ash_ it set the folder as /System/Library/Frameworks/OpenGL.framework/OpenGL but it should be: /System/Library/Frameworks/​OpenGL.framework/Libraries/
20:22 japhb Huh.
20:22 japhb That sounds like a version-dependent thing
20:22 japhb OK, I guess we'll have to detect that somehow.
20:23 japhb Have you got the Perl and C skills to fix the configure stage and make sure it works with both old and new locations?
20:23 japhb (Oh, and/or the Makefile.in, come to think of it)
20:23 ash_ sure, but i don't know the old layout
20:24 bacek o hai
20:25 japhb I would naively assume the old layout is what it was set to before ... but we need another Mac user to confirm.
20:25 mikehh hi bacek
20:25 bacek mikehh, thanks for work on boehm_gc branch
20:26 mikehh bacek: mostly getting codetest to pass - still needs work elsewhere
20:30 dalek parrot: r43328 | bacek++ | branches/boehm_gc/src/gc/gc_boehm.c:
20:30 dalek parrot: Call Parrot_pmc_destroy in finalizer cb.
20:30 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43328/
20:32 ash_ japhb: alternatively you can try linking against /usr/X11/lib/libGL.dylib which will open with X instead of Quartz
20:40 szbalint joined #parrot
20:43 GeJ Good morning everyone
20:43 Coke ~~
20:44 Tene Hi, GeJ!
20:46 GeJ Heya Tene, Coke.
20:51 lucian joined #parrot
21:22 ash_ so... adjusting all the load paths in runtime/parrot/opengl.pir it loads fine
21:22 ash_ but now when it calls one of the glut methods i get: GLUT: Warning in triangle.pir: The following is a new check for GLUT 3.0; update your code.
21:22 ash_ so it becomes loadable but not callable, hmmm
21:30 bacek joined #parrot
21:47 mikehh boehm_gc branch: All tests PASS (pre/post config, make corevm/make coretest, test, fulltest) at r43328 - Ubuntu 9.10 amd64 (g++ with --optimize)
21:53 mikehh bacek: looks good - do we have any benchmarks to check different GC variants
22:04 eternaleye joined #parrot
22:14 japhb ash_, "GLUT: Warning in triangle.pir: The following is a new check for GLUT 3.0; update your code."  UR?  Can you show the complete error?  "The following" seems to indicate it was about to say something else ....
22:14 ash_ GLUT: Warning in shapes.pir: The following is a new check for GLUT 3.0; update your code.
22:14 ash_ GLUT: Fatal Error in shapes.pir: redisplay needed for window 1, but no display callback.
22:15 japhb ?!?
22:15 japhb OK, that's just WEIRD
22:15 japhb problem building libglutcb during parrot build?
22:15 japhb (Or maybe that's got the same path issues you found earlier?)
22:16 ash_ it happens when you call glutMainLoop()
22:16 ash_ googling the error saids its because glutDisplayFunc wasn't getting set, but it is
22:16 mikehh pmc_freeze_cleanup branch: All tests PASS (pre/post config, make corevm/make coretest, test, fulltest) at r43328 - Ubuntu 9.10 amd64 (g++ with --optimize)
22:17 japhb ash_, yeah, I'm thinking something's hinky with libglutcb, which is the magic bit of Parrot code that allows GLUT callbacks to work with Parrot NCI
22:19 ash_ how is libglutcb made?
22:21 ash_ ah, i see, its generated, from config/gen/opengl.pm i'll look at that and see if i can find any problems with it
22:23 japhb ash_, excellent, thank you.
22:23 japhb That was always basically a hack around limitations of existing Parrot NCI
22:23 japhb It's a minor miracle it works at all.  :-)
22:32 ash_ japhb: so... glutDisplayFunc takes a C function pointer, someone its being called in parrot with a parrot sub instead, isn't the point of glutcbDisplayFunc to wrapper glutDisplayFunc to take a parrot sub and store it then pass a C function that calls the parrot one? i think, if i am understanding glutcbDisplayFunc right thats what it should do. Is there a reason that its not being used?
22:32 ash_ s/someone/somehow/
22:37 japhb ash_, when the glutcb* functions are imported into another namespace, they are renamed to get rid of the 'cb', so the user does not need to know about the hack.
22:37 ash_ oh, okay, i understand
22:37 japhb See lines 289 through 303 of OpenGL.pir
22:37 japhb ah, good.
22:40 ash_ can you start parrot with the gdb?
22:43 japhb ash_, In general yes.  Also, I could, but I'm not sure how that would help here, as I don't have a mac dev box.  :-)
22:44 ash_ well i think the issue is in some of the C code and i can't view that with --trace :P
22:46 japhb Grr, we used to have purl trained to know who the Mac users were around here, but I've forgotten the write factoid key, or purl has.
22:48 ash_ gdb is a linux thing too
22:48 ash_ they'd know
22:50 japhb ash_, right ... but I'm saying "Everything works in Linux.  Last I heard, it works on Mac OS X 10.5 and earlier.  It fails on Mac OS X 10.6.  So me doing a gdb on my Linux box will not help find the problem on your OS X 10.6 box.  :-)"
22:50 ash_ ah, i found someone in the perl6 channel that said its failing on 10.5 too
22:50 ash_ same issue
22:50 purl same issue is in Formatter::Youtube
22:50 eternaleye joined #parrot
22:51 japhb Ah, that's getting a little warmer.  At least we know it *used* to work on 10.5.
22:51 ash_ i can't find anyone (online) with 10.4 though to check that one
22:51 japhb nodnod
22:53 japhb Infinoid, ping
22:58 bacek mikehh, It actually not quite good... When I switch default GC to Boehm it crashes on building PGE...
22:58 Whiteknight joined #parrot
23:00 dalek TT #1400 created by jth++: 1.9.0 Compilation on Mac
23:22 Infinoid japhb: pong
23:24 japhb Infinoid, Was trying to help ash_, but don't know the OS X world.  I noticed you were (long ago) in the section of OpenGL.pir that sets paths for OpenGL / GLUT libraries.  Can you lend any insight to ash_'s problem?
23:25 Infinoid I don't know the OS X world, either.  I'm a linux/x64 geek
23:26 ash_ i am not exactly sure how to figure out where the next problem is to be honest, i think its in glutDisplayFunc but i can't tell for sure
23:26 japhb Ah.  You were on the svn blame for the OS X lines
23:27 ash_ can i add breakpoints in the gdb to a extension? (specifically the /runtime/parrot/dynext/libglutcb)
23:27 Infinoid japhb: Was that svn blame from r27022?
23:27 Infinoid r27022: Apply (slightly modified) OpenGL patch from Geoff Broadwell in RT #52988.  Geoff++
23:27 Infinoid :)
23:27 japhb No, a later one.  Checking
23:28 japhb 27059
23:28 * japhb laughs
23:29 ash_ my gdb foo is weak
23:29 japhb That was a patch as well -- you committed, I formatted the patch, the original appears to have been from tetragaon/kid51/ambs
23:29 japhb OK, so I get the "well, duh" award for the afternoon.
23:29 Infinoid Yeah.  I think I basically just let them kick it around until they were all happy, and then applied it blindly
23:29 Infinoid Sorry... :)
23:30 japhb tetragon, kid51, ambs: If any of you are available, we need some Mac OS X skillz to help ash_
23:30 japhb Infinoid, np, sorry to have bothered you without checking the log entry first.  ;-)
23:31 * japhb almost forgot there was a time I needed to have someone else apply patches for me.  ;-)
23:31 Infinoid a very short period of time... :)
23:37 ash_ why do you load a dynext?
23:42 ash_ nm, i figured it out
23:50 ash_ well, i gotta head home, i'll work on this opengl thing a bit more if i have time

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

Parrot | source cross referenced