goyox86 has quit [Read error: Connection reset by peer]
Bwild has quit [Ping timeout: 256 seconds]
goyox86 has joined #rubinius
goyox86 has quit [Ping timeout: 245 seconds]
goyox86 has joined #rubinius
max96at is now known as max96at|off
[spoiler] has quit [Ping timeout: 255 seconds]
meh` has quit [Ping timeout: 240 seconds]
Bwild has joined #rubinius
[spoiler] has joined #rubinius
goyox86 has quit [Read error: Connection reset by peer]
goyox86 has joined #rubinius
havenwood has joined #rubinius
[spoiler] has quit [Quit: Leaving]
unreal has joined #rubinius
goyox86 has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
havenwood has quit [Remote host closed the connection]
sshao has joined #rubinius
Akanksha08 has joined #rubinius
sshao has quit [Remote host closed the connection]
sshao has joined #rubinius
sshao has quit [Ping timeout: 245 seconds]
|jemc| has quit [Quit: WeeChat 1.0.1]
Bwild has quit [Ping timeout: 255 seconds]
amclain has quit [Quit: Leaving]
noop has joined #rubinius
<yorickpeterse>
morning
benlovell has joined #rubinius
benlovell has quit [Ping timeout: 245 seconds]
max96at|off is now known as max96at
benlovell has joined #rubinius
Akanksha08 has quit [Quit: Page closed]
benlovell has quit [Ping timeout: 255 seconds]
goyox86 has joined #rubinius
noop has quit [Ping timeout: 255 seconds]
noop has joined #rubinius
noop has quit [Ping timeout: 245 seconds]
noop has joined #rubinius
nirvdrum has joined #rubinius
<yorickpeterse>
brixen: any plans on a new release for the 1.x branch? I'll actually look into potentially running this instead of REE, but I see you and sshao made a bunch of JIT changes and such since
<yorickpeterse>
which would be nice to have I suppose
<brixen>
|jemc|: I'm going to make configure search a few places for llvm-config
<brixen>
|jemc|: the problem on os x is that brew installs llvm keg only
<brixen>
which is somewhat reasonable but also a pain
<brixen>
the #1 thing we need to fix is depending on any ruby to build rbx releases
<brixen>
the build system needs to have explicitly different systems for dev vs release / pkging
<brixen>
that mean rewriting configure in simple sh
<brixen>
anyone can get started on that :)
<brixen>
please push to a public branch so I don't waste time redoing effort
<|jemc|>
brixen: to clarify: `sh` and not `bash`?
<brixen>
also, it would make sense to make even dev building not require Ruby
<brixen>
and have a specific 1. bootstrap, 2. dev, 3. release build segregation
<brixen>
|jemc|: we should probably try for sh
<brixen>
people use zsh
<brixen>
you can probably follow chruby's lead in the shell you write
<brixen>
but they've had lots of issues
<brixen>
even with bash 3 vs 4 features
<brixen>
lowest common denominator, simplest possible code
<brixen>
that's the best
<brixen>
needs to just work
<|jemc|>
heh, I agree it's best, but every time I try for sh I get frustrated and use bash features :P
<brixen>
it needs these phases: 1. platform reqs, 2. pkg reqs, 3. libc reqs, 4. rbx options (eg bundled libyaml)
<|jemc|>
or these days, when it's just for me I just use fish instead :)
<brixen>
yeah, it's tempting
<brixen>
that's why we've used Ruby for so long
<brixen>
but it's the biggest problem with getting rbx in front of people
<brixen>
and that problem really impairs our ability to deliver
mbj has joined #rubinius
<brixen>
in an internal event with a bunch of people, they basically divided into everyone who had rbx vi ruby-install and the 5 people who wanted to use rbenv
<|jemc|>
yeah, it'll be especially important for titanius to be ruby-independent
<brixen>
(which is fine, that's their choice)
<brixen>
30 min later with multiple people helping, including me, those 5 people still didn't have rbx running
<brixen>
it's just stupid difficult
<brixen>
s/vi/via/
<|jemc|>
rbenv runs into the binstubs issue if you install from source
<brixen>
yep
<brixen>
and others
<brixen>
anyway, it's extremely high priority to get rid of ruby dep to build rbx releases
<brixen>
I guess I need to spend a couple days implementing Symbol GC too
<|jemc|>
brixen: I take it you don't want to try to use autoconf to generate the configure script?
<brixen>
nope nope nope nope
<brixen>
:)
<yorickpeterse>
brixen: melbourne still takes a shit
<yorickpeterse>
blegh, multi-versions on Arch is a total pain
<brixen>
lol, you merged that PR yorickpeterse :P
<yorickpeterse>
I'll take a look at this monday, I have to run in a few minutes
<brixen>
ok
<yorickpeterse>
brixen: Yeah my brain is like cheese
<brixen>
no worries
<brixen>
mine is too, through all those holes I catch glimpses of past events
<brixen>
anyway
<brixen>
I need to remove generating the grammar and fix those issues for bison 3, obviously
mustmodify has joined #rubinius
<brixen>
again, we need to remove this from the basic dev build
<brixen>
no one needs to care about this crap unless you're actually working on it
<brixen>
but it impacts everyone constantly
<mustmodify>
Strange question: does anyone know if there is a ruby function that will check a file to see if it's lexically parsable as ruby?
<mustmodify>
kinda like ruby -c but via Ruby?
<brixen>
nothing built-in in MRI that I know of
<mustmodify>
oh now I feel like I missed something important. What needs to be removed from the dev build?
<brixen>
you can use the parser directly in rbx to do that
<brixen>
mustmodify: the build system needs to support 3 distinct uses: 1. bootstrapping (eg building the parser, compiler initially), 2. normal dev building to mod rbx, 3. release / pkg building
<brixen>
uses 2 and 3 shouldn't depend on a Ruby
<mustmodify>
sure.
diegoviola has left #rubinius ["WeeChat 1.0.1"]
<mustmodify>
Looking at the rubinius-melbourne repo, it seems like if I'm in RBX I need to require 'rubinius/melbourne/melbourne' to access Melbourne... but that doesn't work. How do I get access to it?
<brixen>
mustmodify: so, one thing to think about is what a simpler API to this would look like
<brixen>
mustmodify: that module nesting is necessary to support the rbx runtime compiler separate from eg the build compiler
<brixen>
or from some other language's compiler
<brixen>
but that makes it more cumbersome to use the tools
goyox86 has joined #rubinius
<mustmodify>
I'm sure there are a lot of pieces at play. Clearly I'm out of my depth here, but couldn't you do something like RubyParser = Rubinius::ToolSets::Runtime::Melbourne
<mustmodify>
the trick would then be to encourage every version of Ruby to support the RubyParser interface. :)
<brixen>
yeah, since Ruby has no concept of actual modules (vs Modules), this is pretty hard to do well