Da
Show enters and exits. Hide enters and exits.
| 00:39:33 | phil | While trying to build (did configure with llvm enabled) I get: |
| 00:40:00 | phil | undefined method `rake_system' for #<Object:0xb7de6950> |
| 00:40:50 | phil | Any ideas? |
| 00:41:48 | rue | At which point? |
| 00:42:06 | rue | Wrong interpreter getting used maybe? |
| 00:42:13 | phil | Building Melbourne for MRI |
| 00:42:13 | phil | CXX bstrlib.c |
| 00:42:13 | phil | rake aborted! |
| 00:42:30 | rue | Try with --trace |
| 00:42:51 | phil | ...trying... |
| 00:43:14 | rue | I have to go tend my knee, sadly, but see how much info you can get (poke in any compiler .out/.log files) and so on and gist/pastie that along with platform info |
| 00:43:30 | rue | This hour might be a bit slow but it will be there for later if no-one happens by immediately |
| 00:49:44 | phil | Filed it as an issue here: http://github.com/evanphx/rubinius/issues/#issue/113 |
| 10:33:41 | rue | Ah, yes, there seems to be a problem with that |
| 10:36:48 | rue | #rake_system is/should be private |
| 10:52:08 | rue | Although something seems off |
| 12:28:27 | boyscout | Not only setup segv_handler when on linux, but if execinfo.h is available - 8a21d50 - Dirkjan Bussink |
| 12:29:25 | rue | Heh, the word order is slightly off |
| 12:31:21 | boyscout | CI: 8a21d50 success. 3005 files, 11489 examples, 35634 expectations, 0 failures, 0 errors |
| 12:48:55 | boyscout | Move Debugger under Rubinius namespace - cfb2726 - Adam Gardiner |
| 12:51:42 | boyscout | CI: cfb2726 success. 3005 files, 11489 examples, 35634 expectations, 0 failures, 0 errors |
| 19:57:13 | dwaite | good morning |
| 20:02:08 | rue | Now you are just pushing it |
| 20:08:09 | dwaite | hehe |
| 20:08:19 | dwaite | well, it is always morning someplace right? |
| 20:12:24 | rue | I suppose |
| 20:12:34 | rue | It is morning in Hawaii now |
| 20:12:38 | rue | Or so |
| 21:41:29 | renchap | Hi. I am trying to build rubinius on macosx 10.6.2 with llvm enabled, and i got this error : http://pastie.org/730862 |
| 21:48:44 | rue | renchap: rm melbourne.bundle if it exists |
| 21:49:22 | renchap | done |
| 21:49:26 | renchap | and still the same error |
| 21:50:04 | rue | Is the file back? |
| 21:50:34 | renchap | yes |
| 21:50:44 | renchap | lib/ext/melbourne/ruby/melbourne.bundle: Mach-O 64-bit bundle x86_64 |
| 21:51:00 | rue | It should not build quite that fast... |
| 21:52:24 | rue | Do a distclean and then try $ rake extensions:melbourne_mri |
| 21:53:03 | renchap | lets tru |
| 21:53:05 | renchap | *try |
| 21:53:09 | rue | It could be your MRI is trying to run 32-bit but it should take longer to build the bundle than that so it may be getting stale data |
| 21:53:17 | renchap | seems it worked |
| 21:53:27 | rue | Cool, stale .o files maybe |
| 21:53:39 | renchap | running rake build again |
| 21:59:55 | renchap | rue: same error :( |
| 22:00:55 | rue | When you ran the Melbourne command, it gave you a couple CXX lines and left you with the .bundle, which seems to be a 64bit one, then you get an error after that when running build? |
| 22:01:21 | renchap | yes |
| 22:01:27 | renchap | and when i run file on the bundle : |
| 22:01:33 | renchap | /Users/renchap/dev/rubinius-2/lib/ext/melbourne/ruby/melbourne.bundle: Mach-O 64-bit bundle x86_64 |
| 22:01:35 | rue | That would be suggestive of your MRI maybe running in 32-bit mode, I think |
| 22:01:53 | renchap | ruby 1.8.7 (2008-08-11 patchlevel 72) [i686-darwin9] |
| 22:01:59 | renchap | installed from macports |
| 22:02:57 | rue | What happens if you try to run in straight with MRI? $ ruby -rpath/to/melbourne[.bundle] |
| 22:05:05 | renchap | same error with macports ruby |
| 22:05:12 | renchap | /Users/renchap/dev/rubinius-2/lib/ext/melbourne/ruby/melbourne.bundle: uninitialized constant Rubinius (NameError) |
| 22:05:15 | renchap | with macosx one |
| 22:05:21 | renchap | seems my macports ruby is broken :) |
| 22:07:38 | rue | No, that error is fine (as such), it seems that it can load the fine just fine |
| 22:07:44 | rue | File |
| 22:08:06 | renchap | yes |
| 22:08:25 | renchap | rake build runs fine with the ruby bundled with osx |
| 22:09:15 | rue | Ah, I misread what you said |
| 22:09:32 | renchap | rubinius 1.0.0-rc1 (1.8.7 cfb2726d 2009-11-25 JI) [x86_64-apple-darwin10.2.0] |
| 22:09:35 | renchap | nice :) |
| 22:09:40 | rue | Yes, sounds like you need to get rid of the MP one |
| 22:09:54 | rue | Use rvm for MRI, ends up easier anyway :) |
| 22:10:21 | rue | If you want to develop Rubinius then you probably do not want to install rbx via rvm, but it is nice for the various others |
| 22:12:11 | dbussink | rue: rvm also builds 32 bit by default which can cause probably the same issue :) |
| 22:12:31 | rue | Can use arch |
| 22:13:00 | rue | rvm should be building whatever is the system default |
| 22:13:29 | renchap | rvm seems to be nice |
| 22:13:47 | rue | That is should be as in should, I dunno if it in fact does |
| 22:14:05 | dbussink | rue: in fact it builds 32 bit :) |
| 23:08:23 | wayneeseguin | dbussink / rue btw that documentation is outdated, rvm looks at the value of sysctl hw.cpu64bit_capable |
| 23:16:26 | rue | Yeps, caught that when I was poking around |
| 23:19:46 | wayneeseguin | Although if it's better to just go hands off I'm definitely ok with that ;) |
| 23:19:53 | wayneeseguin | Far less work |
| 23:23:13 | rue | By and large it would be, but I am not really confident that all OS X users are quite aware of their environment nuances |
| 23:23:21 | rue | So long as the default is overridable |