<headius>
when I duplicated the CRuby logic there were some things I was unclear on, like how it shifts back and forth to UTC during zone negotiation
<headius>
I think this is closer to the original logic
<headius>
passes specs locally, but we'll see how it does in CI
<headius>
this form of Time.at does not appear to be well-tested
<headius>
ok ttfn, back in the morning
razetime has joined #jruby
<yardenlaif[m]>
headius: The local variable name change would really simplify our variable parsing code, and also help out existing Java debuggers work with JRuby
<yardenlaif[m]>
Other than that Java debuggers require the compile mode to be FORCE
<yardenlaif[m]>
From what I can tell I can't programmatically force the compile mode to be FORCE
razetime has quit [Ping timeout: 260 seconds]
razetime has joined #jruby
genpaku has quit [Remote host closed the connection]
genpaku has joined #jruby
sagax has joined #jruby
razetime has quit [Ping timeout: 268 seconds]
razetime has joined #jruby
<headius>
There's a property you can set but I'm not sure if we check it again after boot
<headius>
So it would be nice if you could switch to Force mode after startup
<headius>
darn almost passed
razetime has quit [Ping timeout: 264 seconds]
razetime has joined #jruby
Guimauve has joined #jruby
<headius>
interesting, without my fix we actually pass all tests on the tzinfo gem
<headius>
but jeremyevans's case fails
<headius>
ok I pushed a more localized fix that doesn't regress tzinfo tests... hopefully sequel will also be happy with it
<Guimauve>
headius - that you're still here and still going strong all these years later is unexpectedly comforting
subbu has joined #jruby
<headius>
enebo: looks like my smaller fix avoided regressions and still solves the issue so I'll merge that in
<headius>
Waiting to hear back about the socket binding thing since I could not reproduce locally
<enebo[m]>
ok small is good
<headius>
The logic for this is so hard to follow in MRI
<headius>
Weird issue just came in for require_relative, I'll take a poke at it
razetime has quit [Remote host closed the connection]
Guimauve is now known as StayPuftGuimauve
StayPuftGuimauve has quit [Quit: Client closed]
Guimauve has joined #jruby
subbu has quit [Quit: Leaving]
Guimauve has quit [Quit: Client closed]
Guimauve has joined #jruby
subbu has joined #jruby
<byteit101[m]>
I suppose I could just use gcc-arm-linux-gnueabi cross compiler for linux?
<headius>
You could look at the GHA builds for JFFI, they use docker and qemu to build Linux binaries across several platforms
<headius>
There are also cross compiler options but I've never managed to get them to work very well
<enebo[m]>
so 19F/E still but they are generally error handling and probably minor
<enebo[m]>
I need to test rubocop or maybe parser? to see if we are still passing everything there but this will probably get merged on Monday
<enebo[m]>
This ended up being a hellish amount of debugging work to combine two type systems into one file but it should make future updates be less overall work
<enebo[m]>
Assuming this ends up not leading to immediate followup work I think I will implement in-order execution of expressions (Ruby 3 feature) early next week. That should be one of our last outstanding Ruby 3 features we need for release (at least in regards to significant language-level features)
<enebo[m]>
I do think we need some work on refinements still but I am not really sure
<headius>
Yeah but two things I know that are outstanding is some modified hierarchy logic for prepend, that was a fix to the change in 3.0, and there are some refinement changes we need to catch up with
<headius>
s/but/the/
<headius>
I filed an issue for the prepend thing
subbu has quit [Ping timeout: 248 seconds]
<headius>
Not exactly sure what's needed for refinements but I know there's stuff in there
<headius>
And of course there's a ton of stuff marked for 9.4 so we'll need to audit that and see what would be fine to push to a .1 release
<byteit101[m]>
hmm... I think I will hope that dockcross and ldywicki can help me out
<byteit101[m]>
and move on to jar packaging in the meantime
subbu has joined #jruby
<byteit101[m]>
oh, do we have a listing of the shell exec commands?
<byteit101[m]>
is it always ["sh", "-c", single_argument_command] ? or are there other cases?
<byteit101[m]>
also, what's a fast way to iterate and test the complete jar?
<byteit101[m]>
I have to use mvn -Pcomplete clean package or else I get errors
<byteit101[m]>
Aww, -Pcomplete doesn't know how to deal with symlinks. got a strange error about "invalid regex options", because /home/byteit101/... looks like a regex instead of the symlink file :-D
<headius>
We have a CI job that runs with the complete jar, but it's a complicated command line
<byteit101[m]>
faster?
<byteit101[m]>
locally it takes me 1:30 to rebuild the jar, I'm hoping to trim that down
subbu has quit [Quit: Leaving]
<byteit101[m]>
headius: Is there a jruby-wide temp dir?
<headius>
You mean a property or environment variable or something?
<headius>
Pretty sure we look at the standard Unix environment variables for temp location, or the JVM does and we just follow that
<byteit101[m]>
I mean for extracting so's to
<byteit101[m]>
I am creating my own temp dir right now, but if there is a system wide one that will be deleted on exit, that would be great
<byteit101[m]>
I see /tmp/jruby-1155459 /tmp/jruby-635075 /tmp/jruby-635491 on my system, with jline, readline, and wait.jar