Index

Show enters and exits. Hide enters and exits.

03:38:54postmoderncould i ask someone to double check a bundler related bug
03:39:02postmodernto see if it's a problem with rubinius, and not bundler
03:39:04postmodernhttp://github.com/carlhuda/bundler/issues/issue/599
04:06:10brixenwhy is rbx running gems from 1.9.2?
04:56:34brixenpostmodern: I commented on your ticket
04:57:11postmodernbrixen, just saw that
04:57:23postmodernbrixen, you should remember, that i *never* install Rubinius using RVM
04:57:44postmodernbrixen, i always compile against head, and just add it to my $PATH variable in my .bashrc
04:58:56brixenwell, something is wrong with it, rbx should never be running gems in some other Ruby's install
04:59:30brixenI just have rbx in my bin dir that links to rbx in my build dir
04:59:35brixenworked fine with that setup
04:59:54postmodernbrixen, ok, so how do you install rubygems for Rubinius?
05:00:02postmodernbrixen, i do not have a system ruby, just RVM
05:00:17postmodernbrixen, so whats the best way to have a self-contained Rubinius install
05:01:20brixenjust clone the repo, build it, and ln bin/rbx somewhere
05:01:25brixenthat's all I do
05:01:38brixenyou don't install rubygems for rbx, it comes with rubygems
05:01:48brixento install a gem: rbx gem install foo
05:02:01postmodernbrixen, but where is the gem command?
05:02:06brixenin rbx
05:02:11brixenrbx gem install
05:02:42postmodernah ha
05:02:56brixentechnically it's in lib/bin, but you don't need to know that ;)
05:03:10postmodernlib/bin ?
05:03:25postmoderni can hear the unix designers rolling in their graves :)
05:03:34brixenrbx does not litter gem binaries around your directories
05:03:46brixenthey are all under rbx, and you use rbx -S foo to access them
05:05:04brixenthe unix designers weren't that smart when it comes to fs standards
05:05:29brixenwitness trying to uninstall any program no installed with a pkg manager
05:05:53postmodernbrixen, maybe it should be opt/lib/bin
05:06:30brixenthe relationship between unix and pkg managers reminds me of the relationship between windows and anti-virus programs
05:10:13brixenpostmodern: I noticed some of your specs are missing "should ..." :P
05:10:33brixendon't you ever get tired of all that redundant noise?
05:12:15postmodernbrixen, muscle memory
05:12:43postmodernbrixen, ok it looks like there's some issues with running rspec2 under rubinius
05:12:52postmodernbrixen, it's actually trying to run it's own bundler command
05:13:27brixenhmm, what is "it's" in that? rspec2 or rbx?
05:13:44postmodernbrixen, http://pastie.org/1105475
05:13:50postmodernbrixen, yes, this is another project
05:14:16postmodernbrixen, originally was suggested to try to catch this infinite recursion, by getting a stacktrace on rubinius
05:14:22postmodernbrixen, since MRI truncates their traces
05:14:38brixenso, what do I do to repro your setup?
05:16:27postmodernbrixen, http://pastie.org/1105480
05:19:46postmodernbrixen, also i think a while back i used can/may/must in spec descriptions
05:20:18brixenwow, look at all these gems...
05:20:34brixenpostmodern: that sounds ridiculous :P
05:20:49brixenwtf is a spec that says "foo may do X"?
05:22:02brixenso, are you just shelling out to "bundle blah" here?
05:22:16brixenyou need to run "rbx -S bundle blah" if you're going to use rbx
05:22:18postmodernbrixen, it appears rspec2 is just shelling out
05:22:28postmodernbrixen, actually i do not
05:22:29brixenwell, who is telling it what to run?
05:22:34postmodernbrixen, the Rakefile is setup to activate Bundler
05:22:59postmodernbrixen, also spec/spec_helper.rb
05:23:08brixenok, well, I don't know all the details
05:23:18postmodernbrixen, magic behind the scenes
05:23:22brixenbut I can tell you that something is assuming it can just exec "bundle"
05:23:31brixenand that assumption is false
05:23:36brixennot rbx's problem
05:23:43postmodernbrixen, well it works under MRI and JRuby
05:23:50brixendoesn't matter
05:24:06postmodernbrixen, so im guessing ENV isn't being inherited properly by the shell command
05:24:52brixenI have no executable "bundle"
05:25:06brixenif you want to run bundle under rbx, you need to exec rbx -S bundle
05:25:14postmodernah right
05:25:28postmodernis there anyway i could add Rubinius's bin/ dir of gems to my $PATH
05:25:36postmodernso i could run them transparently
05:25:58postmodernor setup stubs to run rbx -S $*
05:26:19postmodernotherwise i cannot test rspec2 apps with rubinius
05:27:52brixenwhy can't youL
05:27:55brixen?
05:28:19brixenif it's an issue with rspec2, we should submit a patch
05:29:00postmodernbrixen, what would be a good way to detect rubinius
05:29:11postmodernbrixen, to somehow prefix the command with rbx -S
05:29:13brixenRUBY_ENGINE == 'rbx'
05:29:24postmodernbrixen, bingo
05:30:34brixenwhat rspec should be doing is getting the ruby exe name and *always* running 'ruby_exe -S bundle'
05:30:44brixenrake has a ruby() function
05:31:59postmodernbrixen, forking
05:32:08postmodernbrixen, then will engage in the usual developer knife fight
05:32:21postmodernbrixen, in arguing whether such a patch is necessary or if it's even a bug
05:32:42brixen1.9 has this very nice... ruby1.9 -rrbconfig -e 'p RbConfig.ruby'
05:32:49brixenwe could add that to rbx
05:33:04brixenpostmodern: I assure you, it's a bug
05:33:21brixenthe ruby_exe -S foo will work across every ruby
05:33:24postmodernapparently they're just systeming it
05:33:36brixenit's absolutey a bug to assume the gem exe is on the path
05:33:50postmodernso s/system/ruby/g ?
05:33:59brixenyes, system("#{ruby_exe} -S foo")
05:34:17brixenwell, if they have access to Rake's ruby command, that's one option
05:34:34brixenit's easy to find out which ruby is running from rbconfig
05:34:39postmodernbrixen, they do, it inherites TaskLib
05:34:44brixenit's kinda messy pre 1.9, but possible
05:34:57brixenin 1.9, it's RbConfig.ruby
05:35:29brixenok, if they have Rake's ruby, should work
05:35:33brixenyou could check that
05:35:38dkubbwould Config::CONFIG['RUBY_INSTALL_NAME'] work to get the ruby exe name?
05:35:49brixendkubb: no
05:36:10brixenyou have to check the ext too for platforms like windows
05:36:23dkubbah
05:36:40brixenthis is what we have in mspec: http://gist.github.com/541828
05:37:43brixenbut again, 1.9 wraps it up into a nice RbConfig.ruby
05:37:50brixenwe could totally add that to rbx
05:46:01postmodernbrixen, and we have lift off
05:46:54postmodernbrixen, thanks for that help
05:47:46brixenpostmodern: sweet! n/p
05:52:32postmodernwow i really hate github auto-responders
05:52:45brixenwhy's that?
05:52:49postmodernnot like it's going to make your inbox any cleaner
05:52:50dkubbthe rspec autoresponder?
05:53:05postmodernand it forces me to retype a bug report, and attach links to commits
05:53:07postmodernyeah
05:53:36brixenI have an autoresponder on rubyspec because I don't like to deal with the jumbled messes that come in as pull requests
05:53:44brixenI accept git fp's
05:54:29brixenand even though I doc that in Contributing, people still send me pull requests
05:54:40brixenso they get an auto response :)
05:59:25postmodernyeah that makes sense for something as formal as rubyspec
05:59:37postmodernbut things such as jekyll or rspec
05:59:51postmodernanyways i hope that gets fixed asap
05:59:57postmodernand pushed out into another beta gem
08:15:07headiushiya
08:15:10headiusanyone awake?
08:15:16headiusI'm trying to build melbourne against jruby
08:15:45headiuscompiles ok, but I can't get it to link...the references to libmquark.a and friends
08:53:33dbussinkheadius: i am yeah
08:53:49dbussinkheadius: dunno how useful i am though ;)
08:53:50headiusI think I have it now...extconf.rb doesn't work, so I'm using the Rakefile
08:53:57dbussinkheadius: ah ok
08:54:11dbussinkheadius: i was wondering, how are you guys handling thread safety for c extensions?
08:55:16headiusI'm not sure, but I think it locks around call-outs
08:55:22headiuscexts are a ghetto
08:55:48dbussinkheadius: yeah, we were thinking of a GEL too ;)
08:55:51dbussinkglobal extension lock
08:56:02dbussinkmaybe an extension can register itself as threadsafe in the future
08:56:27headiusyeah, I suppose that might be possible
08:56:41headiusif anyone's using threads with jruby we'll probably just tell them to avoid cexts
08:56:51dbussinkcurrently the hydra branch is pretty thread unsafe with regard to c exts
08:57:03dbussinkavoiding c exts is quite an impossibility with rbx :)
08:58:20headiusyes, that's true
08:58:58headiushmm, looks like I got melbourne to build
09:00:08headiusno changes required, just had to fix a few rbconfig properties
09:00:15dbussinkheadius: ah, cool :)
09:00:35headiuswell, that's another goofy project done for tonight
09:00:55dbussinkheadius: hehe, well, you can try the DO drivers, but i don't know how friendly they are
09:01:00headiushmmm
09:01:04dbussinkheadius: since they already use jdbc for jruby
09:01:16headiusif they build for rbx they'd probably build for jruby
09:01:17dbussinkso it would be mostly an academic exercise :)
09:01:22headiusI think we're getting pretty close
09:01:48headiuswhat repo should I use?
09:01:55dbussinkdatamapper/do
09:17:11headiusdbussink: so what could I try to build in here to test this?
09:18:27dbussinkheadius: do_sqlite3 should be easiest
09:18:37dbussinkbut it uses rake-compiler and will probably detect jruby
09:18:42headiusdoes it bind directly to sqlite3? doesn't use sqlite gem?
09:18:50headiusahh hmm
09:18:57headiushow does rake_compiler work?
09:19:03headiuswrt C eexts
09:19:19dbussinkheadius: binds directly to sqlite3 yeah
09:19:34dbussinkheadius: well, it makes cross compiling for different platforms way easer
09:19:35dbussinkeasier
09:19:35phlebasdbussink: rake-compiler now has jruby support
09:19:46headiusphlebas: hey, hello there :)
09:19:58headiusI've been having fun with cext tonight
09:19:59phlebashi, just got up and read you tweets
09:20:01phlebascrazy
09:20:14headiusmost things haven't required many changes
09:20:25headiusbut then, I haven't actually gotten most of them to *work* either :)
09:20:25dbussinkphlebas: i know, that's why i said it will detect jruby
09:20:37phlebasoh sorry, misunderstood you there
09:20:41headiusdbussink: does rake_compiler eliminate extconf?
09:22:05dbussinkheadius: it still uses extconf during gem installation
09:22:14headiuschecking for sqlite3_open() in -lsqlite3... no
09:22:14headius*** extconf.rb failed ***
09:22:14dbussinkheadius: so it's only a dev time dependency
09:22:21headiusam I missing something?
09:22:38dbussinkheadius: hmm, what's the mkmf.log output?
09:23:35dbussinkheadius: if you run extconf.rb directly, it doesn't use rake-compiler
09:23:39headiusld error, symbol(s) not found
09:23:41dbussinkrake-compiler wraps around that
09:23:49headiusyeah, I'm trying to do what the gem would do
09:23:54dbussinkso you can easy do cross compilation etc.
09:23:58dbussinkhmm, weird
09:24:03dbussinkit should pickup your system sqlite3
09:24:09headiusworks with MRI...strange
09:28:06headiusdbussink: if I got MRI to build, what's a quick way to try it
09:28:07headiuser
09:28:12headiusI mean if I got mysql to build
09:28:25headiuswhich I seem to have
09:28:36dbussinkheadius: you can try to open a connection to a mysql db
09:28:50dbussinkheadius: DataObjects::Connection.new("mysql://root@localhost") or something like that
09:28:53headiusoh, heh, I wonder if I have one of those
09:28:59dbussinkafter you loaded do_mysql
09:30:54headiusheheh
09:30:55headius~/projects/do/do_mysql ➔ jruby -rubygems -I ../data_objects/lib -I/Users/headius/projects/jruby/lib/ruby/site_ruby/1.8/java/do_mysql -e "require 'do_mysql'; DataObjects::Connection.new('mysql://root@localhost')"
09:30:55headiuscalling init (29cb6d32)
09:30:56headiusdyld: lazy symbol binding failed: Symbol not found: _mysql_init
09:30:56headius Referenced from: /Users/headius/projects/jruby/lib/ruby/site_ruby/1.8/java/do_mysql/do_mysql.bundle
09:30:56headius Expected in: dynamic lookup
09:31:03headiuswell, it did build :)
09:42:22dbussinkheadius: hehe :)
09:42:35dbussinkheadius: babysteps :)
09:42:42headiusoh well...at least I know cext stuff is working very well
09:45:03dbussinkhehe, true :)
09:45:13dbussinkbut i'm off for some shopping etc.
09:45:55headiusseeya
17:27:31philohi
19:34:04philobrixen hi
20:49:12dbussinkbrixen: how's your signal knowledge?