Index

Show enters and exits. Hide enters and exits.

03:32:23rueNeither
04:40:22ruebrixen: A method in test/unit gots the stack unbalanced when compiling
04:45:27rue#129
04:47:27rueObviously the error handling
04:53:14rues/Obviously/Apparently/
07:09:15EliasAmaralis rubinius faster than the reference implementation? does it implement all the language?
07:09:29EliasAmaraland where is the faq? =P
07:10:08EliasAmaralhmm. http://github.com/evanphx/rubinius/blob/master/doc/faq.txt , ko, it doesn't run everything
07:10:34EliasAmaralbut it is faster, right? :( what one could expect of it?
07:22:34jammimaybe one day
07:24:29EliasAmaralmaybe one day it will implement the entire ruby?
07:24:38jammimaybe
07:24:47EliasAmaralokay, do you have benchmarks?
07:24:58jammino, nothing recent
07:25:34EliasAmaralbut what one could expect? compared to, say, java with jvm from sun
11:24:49rueEliasAmaral: No, it is not faster than Java. None of the Ruby implementations are
11:25:15EliasAmaralhmm.. but it could be?
11:25:39EliasAmarali heard llvm is quite impressive
11:25:49EliasAmarali know nothing about it myself
11:34:19rueIt is unlikely. Ruby is generally "fast enough". If your use case and profiling finds Ruby too slow by several orders of magnitude, you probably best drop down a level in the near term.
12:47:07jammiyep, "95% of time is spent on running 5% of the code"
12:47:36jammiusually no ruby implementation is quick enough for the slowest parts -> c
19:07:59brixenwhy does linking have to be such a pita
19:32:22rueBecause it is the lifeblood of HTTP?
19:32:27rueNo, wait, that is probably not it
19:37:32ruebrixen: Saw 129, rite?
19:37:47rueI have not played with it any further
19:40:31brixenrue: yes I see it
19:40:43brixenthis is after evan's last commit I'm assuming?
19:41:01brixenwas there an unbalanced stack before that?
19:42:20rueI did not check, it is head
19:42:35Zoxchow do you plan to fix the unbalanced stack?
19:43:20rueA book under one leg was my plan
19:48:05rueZoxc: This is unbalanced stack in the compilation stage, by the by
19:48:41rueDetected
19:49:18ZoxcDetected isn't good enough, it should be fixed :D
19:49:55rueThanks for volunteering :)
19:52:48ZoxcI'm just fixing it in mirb ;D
19:54:01brixenZoxc: why would you have an unbalanced stack here in mirb?
19:54:24Zoxchere = ?
19:54:35brixenI dunno
19:54:48brixenare you saying you have an unbalanced stack in this same test code?
19:54:53ZoxcI know you can have generate unbalanced stacks in both mirb and rubinius
19:55:25Zoxcmirb won't warn you about it though
19:56:00Zoxcor are you talking about something that escaped your stack analysis?
19:56:06Zoxcand where is the test code?
19:56:50brixenit didn't escape the stack analysis
19:57:02brixenthe stack analysis is what gives the unbalanced stack error
19:57:19rueYes, hence my correction about detection
19:57:20brixenevery method's stack depth is calculated
19:57:43brixenevery point the IP can reach from more than one source has to have the same stack depth from all sources
19:58:00brixenthat's what the unbalanced error is saying
19:58:59ZoxcMy plan was to calculate the absolute and relative stack positions of all jumps and possibly pass variable argument using heap memory instead of using the stack if necessary.
20:31:11dwaitegood afternoon
20:38:58brixenhey dwaite
20:49:42dwaitewhats up brix?
20:50:04dwaiteI'm trying to understand the llvm jit better :)
20:52:01dbussinkand another issue submitted :)
20:52:14dbussinkbrixen: http://github.com/evanphx/rubinius/issues/#issue/130
20:52:28dbussinkbrixen: anything i could look at to deeper investigate what's going on there?
20:57:03brixendwaite: sounds fun
20:57:10brixendbussink: I'll take a look after lunch
20:58:40dbussinkbrixen: btw, i was trying to add the compat.h header and test some performance when i bumped into this one :)
20:58:42brixendbussink: make sure you deleted all your .rbc files
20:59:07brixenbbiab..
20:59:26dbussinkbrixen: already checked tat
20:59:27dbussinkthat
21:37:59dwaitebrixen: maybe I should finally finish setting up my blog and write a long post on how the JIT works once I finish?
21:38:07dwaite(try to beat one of you two to the punch)
22:01:41brixendwaite: sounds like a good idea
22:03:02brixenand... it's snowing here
22:03:05brixenlame
22:03:06brixenheh
22:20:00brixendwaite: ping
23:11:44dwaitebrixen: ping timeout
23:19:59rueBump up your ICMP priority
23:19:59brixendwaite: hey, trying to create a dylib so the mkmf discovery code can link against it
23:20:25brixenand I'm getting a symbol not found for _environ and I'm not understanding why
23:20:35brixentrying to google for it is useless
23:21:10brixenI'm basically linking the same as creating the .bundle and I've gotten everything else to resolve
23:21:20brixenie all the external and llvm stuff
23:31:56dwaitehmm environ?
23:32:23brixenyeah, the libc function to access the evn
23:32:27brixener env
23:32:31brixenit's odd
23:33:20dwaitehmm, scroll to the second paragraph in man environ's PROGRAMMING section
23:34:00brixenoh jesus h christ
23:34:01brixenheh
23:34:22wayneeseguinbrixen: evening~
23:34:24dwaitewhelp, my work here is done
23:34:29dwaiterides off into the sunset
23:34:41brixenhey wayneeseguin
23:34:55brixendwaite: thanks
23:35:12dwaitebrixen: no problem, honestly it was a second-set-of-eyes thing :)
23:35:37ruewayneeseguin: Unbalanced stack not yet resolved
23:35:39brixenyeah, I guess I did encounter that in my google travels
23:35:44brixenjust didn't make sense
23:35:53wayneeseguinrue: nope not yet :/
23:36:07rueRight, no ;)
23:36:30rueThough I should try to figure out what the issue with JIT enabled is
23:38:48dwaiteI like how the rvm install help now is nice enough to personalize the "Thank you for using rvm" note
23:39:10rueLittle touches
23:39:12wayneeseguin::grin::
23:39:16wayneeseguinhad fun with that one
23:42:19dwaiteI'd be most impressed if you special-cased a local install such that you weren't talking to yourself
23:43:57wayneeseguinsorry?
23:44:07rueHe talks to himself
23:45:13wayneeseguinWell I do but that's a whole other discussion :-p
23:45:15wayneeseguin::grin::