Index

Show enters and exits. Hide enters and exits.

00:24:08marcandrecezarsa: are you around?
00:42:21marcandreHmmm. From the latest head, I get "vm/capi/bignum.cpp:70: error: ‘isinf’ was not declared in this scope"
00:47:06cezarsamarcandre: yep, now I am
00:47:16marcandreHi!
00:47:17cezarsahmm, that's certainly my fault :/
00:47:29marcandreYeah, it might be a platform issue.
00:47:38cezarsayep, one sec
00:47:45marcandreJust to let you know, I'm on OS X.
00:48:49marcandreI commented the 4 lines and everything else is OK. I tried to #include some stuff to fix that but failed...
00:49:43evanmarcandre: did you include math.h ?
00:49:59cezarsamarcandre: could you try this patch http://pastie.org/488561
00:50:07cezarsaand let me know if it solves the problem
00:50:18marcandrek, 2 secs
00:52:38marcandrecool, that works! Thanks!
00:53:26cezarsanice, I'll push it now
00:58:54cezarsathat's odd, I'm getting a "Permission denied (publickey)." from github
01:03:11boyscout1.8.7: String#partition (implemented new method) - dfc8ac2 - Marc-Andre Lafortune
01:03:11boyscout1.8.7: String#rpartition (implemented new method) - 727bed8 - Marc-Andre Lafortune
01:03:14marcandreWorks for me.
01:04:08boyscoutAdding RFLOAT to capi, with specs. - 61980b9 - Cezar Sa Espinola
01:04:08boyscoutFixing platform issue in capi's Bignum. - e0f5329 - Cezar Sa Espinola
01:05:15cezarsayeah, github servers were just taking a break :)
01:05:30boyscoutCI: 727bed8 success. 2683 files, 10344 examples, 32915 expectations, 0 failures, 0 errors
01:05:40marcandregreat. Thanks for the quick fix.
01:05:46cezarsanp
04:57:01boyscout1.8.7: String#upto (changed for 1.8.7 and 1.9 compatibility) - a315d91 - Marc-Andre Lafortune
04:57:01boyscout1.8.7: Struct#each, #each_pair (now with optional block) - 596eb8e - Marc-Andre Lafortune
04:57:01boyscout1.8.7: Symbol#to_proc (implemented new method) - c11dadf - Marc-Andre Lafortune
04:57:01boyscoutFixed Numeric#step - 190348b - Marc-Andre Lafortune
04:57:01boyscout1.8.7: Numeric#step (now with optional block) - 5bd146a - Marc-Andre Lafortune
05:15:05boyscoutCI: 5bd146a success. 2683 files, 10336 examples, 32894 expectations, 3 failures, 14 errors
05:59:11libclol, jit's visit_stack_swap pushes in invalid order
06:06:40marcandreNot sure how I manage to break rbx. I have no error here...
06:13:48marcandreI can't see how that's related to anything I've changed, and I'm too tired.
06:14:43marcandreApologies in advance if I did break something!
15:48:57balinthi, is there a way to enter into a debugger session when running the mspecs?
15:49:48balintif I compile with "rake build:debug" I can print debug messages but it would be cool to have an interactive debugger session
16:06:47libcevan, here?
16:19:56ruebalint: --gdb option maybe? You can certainly just do `gdb --args vm/vm whatever the invocation is`
16:20:36rueOh, and if you mean the Ruby debugger, there was support for that too, but dunno if all the functionality is back in place yet?
16:21:20balintrue: yes, I meant the ruby-debug actually, but I'll give it a try with the gdb approach
16:21:23ruelibc_: Heh, I see it too
16:22:07ruebalint: You can get the Ruby callstack in GDB with `p __printbt__(callframe)` -- you obviously need to have some callframe object in scope
16:22:22balintrue: so if I want to debug './bin/mspec core/dir' then I should type 'gdb --args vm/vm ./bin/mspec core/dir', is that right?
16:22:51libcrue, heh
16:23:12rueThe names are a bit bad
16:23:47ruebalint: Something like that, yeah, probably want to just give full path to mspec but otherwise
16:25:52ruelibc_: I will push a change, dunno if evan is planning on being around at all today. US holiday
16:26:42libcrue, I'm not done :) JIT is full of issues… wanted to ask about create_block scope thing
16:28:10rueOh, sure--I meant that as two separate statements :)
16:28:31ruelibc_: What else have you run into?
16:29:22libcswap (wrong order), create_block (call_frame->promote copies variable scope, but vars_ is not updated), and then new shiny bug Unable to send 'isatty' on 'Platform::POSIX' (Module)
16:31:18rueHm. Are other POSIX methods working?
16:34:23balintrue: hmmm, so if I want to debug why a certain mspec that tests a function written in Ruby (Dir.glob) fails can I still use gdb?
16:34:56ruebalint: It would be very awkward, to say the least
16:35:18rueWhich one is it that is failing?
16:35:23balintrue: :) that's what I thought
16:36:27rue'Course at some point you might have to drop to GDB if there is a primitive call involved or something, but the Ruby-side debugger would be better (ours, that is, it is different from ruby-debug)
16:36:30balintrue: I have now corrected it by only using debug messages but it felt awkward, too :) it was "Dir.glob recursively matches any nondot subdirectories with '**/'"
16:37:31balintrue: ok, so how does one use the Ruby one?
16:52:13rueTry just -d, I think the method is Kernel#debugger
16:53:05boyscoutFixed JIT visit stack swap to actually swap. - ca8da19 - Eero Saynatkari
16:55:19boyscoutCI: ca8da19 success. 2683 files, 10336 examples, 32894 expectations, 3 failures, 14 errors
17:03:14balintrue: thank you. I did 'ruby -d ./bin/mspec core/dir' and received 'NameError: Missing or uninitialized constant: MethodContext', is this a known thing?
18:34:37rueYes, a known thing, but then you are gone already...
19:30:55libchuh
19:31:00libcgot irb working
20:12:33rueThis is both a joyous and a sad occasion :)
20:16:35ddubrue?
20:18:17rueI am not sure
20:18:50rueI presume that irb had broken at some point :)
20:30:01libcstill doesn't work, I was delusional after fixing meta_op_minus and wrote -Xrbx.jit.enable :)
20:41:09rueHeh, full(ish) JITting irb? What are you running into?
20:58:53libcI got 2 errors now :) One when not compiled without JIT method called on nil, when precompiled it's 0 + -1 = 1073741823
21:47:01dbussinkanyone interested in a nice backtrace with a bus error?
21:49:35rueSure
21:50:56libcok, just for the record, current patchset: http://pastie.org/489353 irb still doesn't work :)
21:55:11dbussinkrue: http://github.com/evanphx/rubinius/issues/#issue/5
21:55:19dbussinki've actually made an issue out of it :)
21:55:32rueOo, Mr. Fancypants
21:57:04rueSo frame 3 has your problem frame
21:59:57dbussinkrue: yeah, question is how it gets there though
22:01:00dbussinkthis is actually a spec that waits for a timeout from the database (test that incorrect password doesn't connect etc), so i guess a gc run interferes with that somehow
22:01:18dbussinkdunno if it's because the extension is unsafe in general or that it exposes a real problem
22:08:20rueThe bad file 0x18 in the funcall could be just random data, seems to pick the correct one next
22:09:24rueBest way to get more data is to hop into that frame and see what the object looks like
22:14:41boyscoutFixing some capi specs. - 8c76628 - Cezar Sa Espinola
22:14:41boyscoutTagging some specs on capi's Bignum. - e67d5d0 - Cezar Sa Espinola
22:17:06boyscoutCI: e67d5d0 success. 2683 files, 10344 examples, 32915 expectations, 0 failures, 0 errors
22:26:13rueMm, I wonder if the const pointer has any unforeseen complications
22:28:55jptixyou want segfaults reported on github instead of lighthouse?
22:29:28rueEverything should move to Github issues eventually
22:30:20jptixi see
22:33:48jptixi'll add one for this http://gist.github.com/117762
22:35:00rueYepr