<GitHub165>
[jruby] enebo closed issue #2688: IOError: stream closed on using same Http Object with use_ssl = true with jruby 1.7.19 https://git.io/pMlO
<GitHub144>
[jruby] enebo closed issue #2556: File::Stat.file? behaves different with CRuby https://git.io/bZTq
<GitHub76>
[jruby] enebo closed issue #2666: Rails console crash when returning from a before_save block in Rails / JumpException$ReturnJump https://git.io/psa4
<GitHub8>
[jruby] enebo closed issue #2555: JRuby 1.7.11 not working in Windows 7 Enterprise version https://git.io/bYwK
drbobbeaty has quit [Ping timeout: 240 seconds]
<donV>
Woohoo! 9.1.9.0 is out!
donV has quit [Quit: donV]
<nirvdrum>
Aethenelle: Thanks again for the help. I finally got the jnr-ffi tests running on Linux/SPARC.
<nirvdrum>
The issue plaguing me for the last day or so was the CPU detection code thought it was 32-bit and dropping half the value of addresses on certain code paths.
<Aethenelle>
that PR done?
<nirvdrum>
It should be. I don't have a 32-bit SPARC system to try (not sure anyone does). Hopefully my ifdef chain is good.
<Aethenelle>
merged... no idea how to publish the jar though...
<nirvdrum>
enebo said they're using the maven release plugin for that.
<nirvdrum>
I can deploy snapshots though.
<nirvdrum>
Actually, I can't.
<nirvdrum>
I might just be missing my maven settings file. This was a new setup.
<nirvdrum>
Yeah, that was it. Sorry for the noise. Snapshot deployed.
<nirvdrum>
But enebo or headius will have to promote to release.
<Aethenelle>
I've never actually dealt w/ deploying stuff to maven...
<nirvdrum>
I just want another set of eyes on this to ensure I haven't screwed it up.
<nirvdrum>
Next up is jnr-posix and jnr-constants.
<nirvdrum>
Then JRuby itself.
<nirvdrum>
Hopefully that covers them all.
<Aethenelle>
all the tests passing for you?
<Aethenelle>
worst case on this, you're missing some types... not at all hard to fix...
<nirvdrum>
Yes.
<nirvdrum>
That's what I've been struggling with for the past 24 hours. But all the tests pass now.
<nirvdrum>
I pushed test fixes directly to master.
<nirvdrum>
Maybe I should've bundled them into the PR. But they needed to be fixed regardless.
<nirvdrum>
As written, the tests didn't pass on Solaris :-/
<nirvdrum>
Aethenelle, headius, enebo: Do any of you have strong feelings (besides backwards compatibility) on how to structure jnr-constants? I'm not sure how it evolved this way, but it doesn't segregate by OS and platform the same way that (J)FFI does.
<Aethenelle>
i have yet to realy touch constants...
<nirvdrum>
It's a handy project. But it seems to have made the assumption of each OS only running on one platform.
<Aethenelle>
...
<Aethenelle>
... there can be only one?
<nirvdrum>
More perplexing, the Rakefile figures out the arch and then ignores it.
<nirvdrum>
I think the loading code supports it, but none of the generated files use it.
<enebo>
nirvdrum: well that sounds weird
<nirvdrum>
If that's the case, maybe that's my answer.
<enebo>
nirvdrum: constants definitely may vary across archs
<enebo>
nirvdrum: at least it makes sense to me
<enebo>
nirvdrum: so I think it sounds reasonable to break up that way
<nirvdrum>
Yes. SPARC aside, I think there's even differences in x86 and x86_64 on Linux.
<enebo>
nirvdrum: perhaps that is best as a PR to see if we have unanimous consent but I bet it was just an overight
<nirvdrum>
enebo: Yeah. I'm not going to introduce anything like this without consensus.
<nirvdrum>
enebo: It would be nice to declare a major new version of some of these projects and break with some of the old. There's duplicate files for the old com.kenai namespace, for example.
<nirvdrum>
Maybe breaking from Java 6 would be nice too.
<nirvdrum>
What's the status on JRuby 1.7 anyway? Is it EOL for real now?