DanielVartanov_ has quit [Remote host closed the connection]
amsi has quit [Quit: Leaving]
meh` has joined #rubinius
josh-k has joined #rubinius
josh-k_ has joined #rubinius
josh-k has quit [Ping timeout: 244 seconds]
tenderlove has quit [Quit: Leaving...]
byprdct has quit [Ping timeout: 264 seconds]
byprdct has joined #rubinius
byprdct_ has joined #rubinius
<jeremyevans> trying to build 2.3.0 with --disable-llvm errors with vm/builtin/jit.cpp:25:31: error: member access into incomplete type 'rubinius::LLVMState'
byprdct has quit [Ping timeout: 272 seconds]
<jeremyevans> trying to build with LLVM errors for me too, but that may be an issue specific to OpenBSD's LLVM port version (which I think is a checkout of LLVM trunk from 02/28)
byprdct_ has quit []
byprdct has joined #rubinius
byprdct has quit []
josh-k_ has quit [Remote host closed the connection]
josh-k has joined #rubinius
<brixen> jeremyevans: why are you trying to disable llvm?
<jeremyevans> originally, I tried building 2.3.0 with LLVM, but as that failed, I tried without
<brixen> with what version of llvm?
<brixen> and how did it fail?
<jeremyevans> As mentioned a few lines above, it's probably due to an issue in OpenBSD's LLVM port. Here's the error message: /usr/local/include/llvm/ADT/DenseMap.h:175:39: error: no member named 'move' in namespace 'std'
<brixen> jeremyevans: my backlog didn't have any other messages
<jeremyevans> However, it should definitely work without LLVM, right? --disable-llvm used to result in a working build, and now it doesn't
<brixen> I almost removed --disable-llvm actually
<brixen> you should not be using it without the JIT
<brixen> I'm more concerned with fixing the llvm issue
<jeremyevans> here's my earlier message: trying to build with LLVM errors for me too, but that may be an issue specific to OpenBSD's LLVM port version (which I think is a checkout of LLVM trunk from 02/28)
<brixen> what version of llvm?
josh-k has quit [Ping timeout: 256 seconds]
<brixen> can you build 3.4?
<jeremyevans> Well, the OpenBSD port isn't really setup for that
<brixen> why would they pull trunk and not the latest release?
<brixen> and they can't pick a version?
<jeremyevans> It either has to use the LLVM version supported by the porting system, or none at all
* brixen consults the googles
<jeremyevans> I believe it was upgraded from 3.4 to a trunk version back in March, to fix bugs that were in 3.4, or get new functionality added since
<brixen> which one of these is a sane openbsd to fetch? http://www.vagrantbox.es/
<jeremyevans> Well, probably OpenBSD 5.5 64-bit + Chef 11.16.0 + Puppet 3.4.2
<jeremyevans> OpenBSD 5.6 is getting released tomorrow. I'm testing on OpenBSD-current
<brixen> so 5.5 is current stable and tomorrow 5.6 will be current stable?
<jeremyevans> Correct
<brixen> and -current is what in relation to 5.6?
<jeremyevans> current is a couple months past that. 5.6 was basically finalized in August
<jeremyevans> You'll want to take a look at the patches here: http://cvsweb.openbsd.org/cgi-bin/cvsweb/ports/lang/rubinius/patches/?hideattic=1#dirlist
<brixen> and what's the typical release schedule?
<jeremyevans> New versions May 1 and November 1
<jeremyevans> Those are the patches used when building 2.2.9
<brixen> is you goal to support 5.6 only or something else?
<brixen> s/you/your/
<jeremyevans> My goal is to have it supported in -current
brixen changed the topic of #rubinius to: 2.3.0 - http://releases.rubini.us/rubinius-2.3.0.tar.bz2 : logs - http://irclog.whitequark.org/rubinius
<brixen> which basically means trunk llvm, which probably is missing features from 3.5 that they want to remove in 3.6
<jeremyevans> 5.6 is fixed, it shipped with 2.2.x
<brixen> like the API we use for the JIT
<brixen> ok
<brixen> that's unfortunate
<brixen> I guess I'll have to build -current from scratch then
<brixen> is there some html somewhere that says how one does that?
<brixen> thanks
<brixen> surprised they still serving that over http
<brixen> they're *
<jeremyevans> Just boot from that. Installer is text-based (a shell script).
RageLtMan has joined #rubinius
<jeremyevans> It'll be fun building from scratch to. You won't be able to use the system compiler with 2.3.0 (GCC 4.2.1), though that worked in 2.2, since you are now using C++11 features
byprdct has joined #rubinius
<brixen> we shouldn't be using them in rbx
<brixen> llvm uses them perhaps
josh-k has joined #rubinius
<jeremyevans> Maybe. I know when I tried with the system compiler, it failed because std=c++11 or whatever wasn't supported, so I switched to a newer compiler
<brixen> if llvm 3.5 is found, it injects that because llvm uses it
<brixen> I've going a patch to fix --disable-llvm
<brixen> I guess it's good I didn't remove that yet
<brixen> but I need to fix rbx to work with 3.6 / trunk
<brixen> unfortunately, I was going to use this hour to finish the homebrew tap for rbx :(
<jeremyevans> brixen: OK, sounds good. Let me know if you'd like me to test a --disable-llvm fix patch
<jeremyevans> homebrew is probably important to more people than OpenBSD
<brixen> jeremyevans: I will push it in 2 min when the specs finish
<brixen> yes, please do test
<jeremyevans> brixen: Great, thanks!
<brixen> it'll take a while for me to get openbsd running
<brixen> n/p, thanks for letting me know
GitHub27 has joined #rubinius
<GitHub27> [rubinius] brixen pushed 1 new commit to master: http://git.io/UMOjow
<GitHub27> rubinius/master f846083 Brian Shirai: Fixed --disable-llvm.
GitHub27 has left #rubinius [#rubinius]
byprdct_ has joined #rubinius
byprdct has quit [Ping timeout: 244 seconds]
<brixen> jeremyevans: I'll probably just work on getting rbx building with 3.6 / trunk and then let you test that
<brixen> setting up openbsd for me is a mega PITA
<brixen> unless someone can make me a vbox machine I can use with vagrant
travis-ci has joined #rubinius
<travis-ci> rubinius/rubinius/master (f846083 - Brian Shirai): The build passed.
travis-ci has left #rubinius [#rubinius]
josh-k has quit [Remote host closed the connection]
josh-k has joined #rubinius
josh-k has quit [Read error: Connection reset by peer]
josh-k has joined #rubinius
GitHub173 has joined #rubinius
<GitHub173> [rubinius] brixen pushed 1 new commit to master: http://git.io/8mhllQ
GitHub173 has left #rubinius [#rubinius]
<GitHub173> rubinius/master 5cd0c42 Brian Shirai: Removed Process.kill spec checking EPERM....
havenwood has joined #rubinius
travis-ci has joined #rubinius
<travis-ci> rubinius/rubinius/master (5cd0c42 - Brian Shirai): The build has errored.
travis-ci has left #rubinius [#rubinius]
<havenwood> grats on 2.3.0!!
<havenwood> and Happy Halloween!
<brixen> havenwood: heh, thanks
<havenwood> Rubinius::Console seems intriguing.
havenwood has quit []
byprdct_ has quit [Read error: Connection reset by peer]
byprdct has joined #rubinius
byprdct_ has joined #rubinius
byprdct has quit [Ping timeout: 245 seconds]
JohnBat26 has joined #rubinius
byprdct_ has quit [Read error: Connection reset by peer]
havenwood has joined #rubinius
<jeremyevans> brixen: I got a lot further with your patch, but I'm still not able to build 2.3.0 without LLVM on OpenBSD. With the system compiler (GCC 4.2.1), I get: error: forming reference to reference type 'rubinius::metrics::metric&' (see http://pastie.org/9689090 for horribly mangled build output)
<jeremyevans> With GCC 4.6 and 4.8, things compile and link, but rbx SIGBUSes when running
byprdct has joined #rubinius
<jeremyevans> With clang, things compile, but don't link.
<jeremyevans> Anyway, if you have ideas on the reference to reference type error, let me know. From what I read, it's not an issue in C++11, but you said that rbx doesn't require that (see http://stackoverflow.com/questions/1155142/why-do-i-get-an-error-in-forming-reference-to-reference-type-map/17627254)
<jeremyevans> brixen: doing some more research, the version of llvm in the OpenBSD ports tree is right before they switched to the LLVM codebase to require C++11. I'm guessing it hasn't yet been updated to 3.5 official release because it would require building a C++ compiler with C++11 support (e.g. GCC 4.8 or 4.9) with the system compiler before being able to build LLVM.
byprdct has quit [Read error: Connection reset by peer]
havenwood has quit [Remote host closed the connection]
meh` has quit [Ping timeout: 272 seconds]
noop has joined #rubinius
josh-k has quit [Remote host closed the connection]
noop has quit [Ping timeout: 256 seconds]
josh-k has joined #rubinius
josh-k_ has joined #rubinius
josh-k has quit [Ping timeout: 245 seconds]
havenwood has joined #rubinius
noop has joined #rubinius
max96at|off is now known as max96at
benlovell has joined #rubinius
Caius has quit [Ping timeout: 260 seconds]
Caius has joined #rubinius
enebo has joined #rubinius
flori_ has quit [Ping timeout: 258 seconds]
flori has joined #rubinius
benlovell has quit [Ping timeout: 244 seconds]
benlovell has joined #rubinius
benlovell has quit [Ping timeout: 265 seconds]
josh-k_ has quit [Remote host closed the connection]
flori has quit [Ping timeout: 245 seconds]
josh-k has joined #rubinius
flori has joined #rubinius
josh-k has quit [Ping timeout: 264 seconds]
havenn has joined #rubinius
havenwood has quit [Ping timeout: 246 seconds]
flori has quit [Remote host closed the connection]
flori has joined #rubinius
<brixen> jeremyevans: hmm, thanks for investigating more
<brixen> jeremyevans: I'd like to get rbx master working on llvm 3.5 + llvm trunk
<brixen> jeremyevans: is it realistic to expect openbsd-current to get working 100% with clang + llvm 3.5/trunk?
<brixen> jeremyevans: pretty sure I'm going to drop support for gcc completely very soon
<jeremyevans> brixen: I'll ask the OpenBSD LLVM port maintainer about upgrading to the final LLVM 3.5 release. I don't think it makes sense to expect other projects like Rubinius to support an LLVM trunk checkout from 8 months ago.
<brixen> jeremyevans: that would be awesome, thank you
<brixen> also, if the maintainer needs info or wants to collaborate, feel free to direct them to us as well
<brixen> jeremyevans: is openbsd planning to do what freebsd did re clang/llvm?
max96at has quit [Read error: Connection reset by peer]
<jeremyevans> brixen: In terms of making it the system compiler, I don't think so, at least I haven't heard anything about it
<jeremyevans> brixen: Many of the large ports that OpenBSD has (chrome/firefox) are built with the llvm port, but the vast majority of the ports compile fine with the system compiler
<brixen> ok, as long as clang is available, recent, and works, we should be ok with rbx
<brixen> can we force use of clang from the rbx pkg?
<brixen> or port
<jeremyevans> brixen: Yep, I was trying that. I can give you the failing linker error, let me try building it again with clang
<brixen> ok
<jeremyevans> I added a -v to the link command so you could the linker command line
<brixen> I'm unsure what the error is
<jeremyevans> Me too
<brixen> did you configure with --cc=clang --cxx=clang++ ?
<jeremyevans> It sets the path up so that gcc/g++ are symlinked to clang/clang++
<jeremyevans> But clang still uses the system linker, since lld is not yet ready (so I read)
diegoviola has joined #rubinius
<brixen> hmm, seems like a mess, unfortunately
<jeremyevans> Yep. Hopefully using LLVM 3.5 release will result in object files that will link.
<brixen> but that may not be till like May 2015?
havenn is now known as havenwood
RageLtMan has quit [Ping timeout: 272 seconds]
<jeremyevans> brixen: In terms of official releases, yes. But OpenBSD users are used to the fact that release doesn't have the most up to date packages.
<brixen> what do they typically do?
<jeremyevans> Just use slightly older packages :)
<brixen> ugh lol
JohnBat26 has quit [Quit: KVIrc 4.3.1 Aria http://www.kvirc.net/]
<jeremyevans> For example, OpenBSD 5.6 ships with Ruby 2.1.2, since that was the current version at the time
<brixen> why don't openbsd users revolt and demand a pkg system that handles updates?
<jeremyevans> The package system handles updates fine
<brixen> so why can't 5.6 update to rbx 2.3+ ?
<jeremyevans> If you upgrade from OpenBSD 5.5 to 5.6, you run pkg_add -u, and all of your packages get upgraded
<jeremyevans> OpenBSD has limited manpower, so all development is done in the current tree. Every six months, that gets branched to do a release
<jeremyevans> Backporting updates from -current to -release is only done for security issues or very serious other issues
<brixen> I understand limited resources but 6 month releases actually cost more effort, IME
<brixen> and time is only accelerating, meaning 6 months is actually getting longer in terms of effort
<jeremyevans> Well, OpenBSD has been doing it for about 18 years this way, and so far it has worked pretty well
<brixen> time is accelerating in terms of the amount of changes happening in a unit of time
<jeremyevans> You shouldn't assume that happens to OpenBSD as well. Change for changes sake is frowned upon there :)
meh` has joined #rubinius
<chrisseaton> brixen: for a paper I'm writing - rbx seems to specifically not inline a call that goes to #method_missing (https://github.com/rubinius/rubinius/blob/master/vm/llvm/inline.cpp#L154-L161) - do you know why this is? is there some kind of trade-off that makes it a bad idea in your experience?
enebo has quit [Quit: enebo]
<brixen> chrisseaton: most things in rbx JIT are a proof of concept evan did in like 3 weeks of work
<brixen> there's massive room for improvement
<brixen> so if something isn't done, it's because we haven't gotten to it yet
<chrisseaton> brixen: ok, thanks very much
<brixen> not because it can't or we don't know how to JIT things
<brixen> which I why I cautioned you against saying "can't"
<brixen> it misrepresents Rubinius and doesn't make you look like you know what you're talking about
<chrisseaton> the only reason I asked about this one is that someone's gone out of their way to exclude it, rather than it being an incomplete feature
<chrisseaton> but maybe some other stuff further down the pipeline isn't expecting method missing to be inlined, or something like that
<brixen> nope, nothing like that
<brixen> if an optimization is possible, rbx JIT will do it
<brixen> probably sooner rather than later, too
<yopp> huh. JIT fight!
<brixen> brew's naming convention for taps is rather odd
<yopp> packageXY?
<brixen> hm, I thought brew removed Rubinius formula
<brixen> yopp: it has to be something like homebrew-foo
<brixen> so we can't just have a rubinius/homebrew repo
<brixen> since brew will expect to find rubinius/homebrew-homebrew
<brixen> I made it rubinius/homebrew-apps
<brixen> so you brew tap rubinius/apps
<brixen> anyway, gotta run
<diegoviola> what is Rubinius::Console, how do I use it?
<cremes> diegoviola: there will be a blog post explaining it. in the meantime, you’ll have to read the code to figure out how to use it.
<diegoviola> ok
<diegoviola> my app seems to work just fine with 2.3, nice work
<diegoviola> even though my app is quite small
<diegoviola> I'll try other apps later
|jemc| has joined #rubinius
RageLtMan has joined #rubinius
|jemc| has quit [Quit: WeeChat 1.0.1]
|jemc| has joined #rubinius
<yorickpeterse> diegoviola: basically Rubinius::Console is a replacement for the agent we had before
<yorickpeterse> it will be a proper REPL with a bunch of other cool features if I remember correctly
<diegoviola> so will irb use that eventually?
<yorickpeterse> No, IRB itself won't use Console
<yorickpeterse> basically you'll be able to use it similar to gdb for Rbx processes
<yorickpeterse> that is, attach to a process, inspect it, run Ruby, etc
<diegoviola> cool
josh-k has joined #rubinius
josh-k has quit [Ping timeout: 245 seconds]
JohnBat26 has joined #rubinius
havenwood has quit [Remote host closed the connection]
noop has quit [Ping timeout: 272 seconds]
houhoulis has joined #rubinius
houhoulis has quit [Remote host closed the connection]
havenwood has joined #rubinius
<heftig> yorickpeterse: hm, rbx console -h just gets me a LoadError
<headius> hmm, I can't build anymore and don't know what I need to do
<headius> seems like I don't have llvm-config or something (OS X Mavericks)
JohnBat26 has quit [Quit: KVIrc 4.3.1 Aria http://www.kvirc.net/]
<|jemc|> headius: do you have llvm?
<headius> I have xcode...I thought that would cover it
<|jemc|> not an OSX guy so I can't say; sorry
<|jemc|> btw I just now figured out what was going on with my czmq on android issue
<havenwood> headius: yeah, OS X doesn't ship with llvm-config
josh-k has joined #rubinius
<headius> |jemc|: oh, what was it?
<|jemc|> used readelf to see that libczmq was not actually linked to libzmq - and sure enough, looking in config.log betrayed that configure did not put -lzmq in LIBS
<havenwood> headius: llvm-config is available in the brew bottle poured for llvm though
<headius> I did just try brew install llvm but configure still failed
<headius> checking log
<|jemc|> however, it only had that problem when compiled with the android ndk
<headius> yeah, still no llvm-config
<havenwood> headius: brew link --force llvm
<|jemc|> I think maybe something is wrong with PKG_CHECK_MODULES in the android ndk - but not sure how I would track it down past there
<havenwood> they don't link llvm by default
<|jemc|> but at least I have a workaround now :)
<headius> |jemc|: so it wasn't build with the dependency, and so it couldn't see across libs
<havenwood> or: $(brew --prefix llvm)/bin/llvm-config
<headius> so there's a possibility then if we wanted to access a specific library with ffi we might have to build ffi with that dep
<|jemc|> yeah, but the configure script works right when I configure for my desktop machine
<headius> ahhh strange
<headius> havenwood: oh, ok
<|jemc|> I should have went to readelf way sooner
<headius> that looks better, thanks
<|jemc|> headius: re FFI: yeah, that sounds like it might be a problem...
<headius> not a huge one, but it would need some infrastructure
<|jemc|> ah, so compiling ffi for each project that uses it, with the dependencies known at that point?
<headius> right...presumably you know what libs you'll need
<headius> and probably a special build of the ffi lib that adds in common system libraries to its own deps
<headius> something like that :-)
<headius> it might also be possible that dlopen(RTLD_GLOBAL) before System.loadLibrary might work too
josh-k_ has joined #rubinius
josh-k has quit [Ping timeout: 258 seconds]
bennyklotz has quit [Remote host closed the connection]
bennyklotz has joined #rubinius
|jemc| has quit [Ping timeout: 258 seconds]
kfpratt has quit [Remote host closed the connection]
<yorickpeterse> heftig: the console command doesn't exist atm
<yorickpeterse> the docs haven't been updated for that
<yorickpeterse> heftig: the console isn't really available yet to userspace, only some basic plumbing is available
<yorickpeterse> most of it is hidden in C++