travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
slyphon has joined #jruby
Puffball has joined #jruby
_whitelogger has joined #jruby
jmalves has joined #jruby
jmalves has quit [Remote host closed the connection]
jmalves has joined #jruby
jmalves has quit [Remote host closed the connection]
jmalves has joined #jruby
<
kares>
headius: hey, yes same here - not having a good understanding for the mutex
<
kares>
+ the COW thing seem to have broken the handle_interrupted? test case you were originally trying to fix
<
kares>
feel free to take the PR branch over ... I might not get back to it till the end of week
Puffball has quit [Remote host closed the connection]
fredmorcos has joined #jruby
fredmorcos has quit [Remote host closed the connection]
nirvdrum has quit [Ping timeout: 250 seconds]
nirvdrum has joined #jruby
lucasb has joined #jruby
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
<
headius>
kares: ah we were hoping to get 9.2.6 out Thursday
<
headius>
Do you know what caused the regression at least? I have not seen a lot of discussion
<
enebo>
kares: possibly weds if we can nail down a couple of things today
<
kares>
headius: its all on the PR and the issue
<
kares>
feel free to take the PR over - there's mostly only cleanup
<
kares>
there's also a test that reproduces the issue fairly well in a few secs
<
kares>
did not dig deeper myself unfortunately
xardion has quit [Remote host closed the connection]
xardion has joined #jruby
Aethenelle has joined #jruby
<
headius>
kares: ok
<
headius>
kares: you believe the PR is done other than cleanup?
<
headius>
kares: I get why the COW fixes
<
headius>
addAll locks on Vector and then tries to iterate the other thread's interrupts
<
headius>
if that other thread also tries to addAll it will be locked waiting on something else
<
headius>
it should have just cloned or something
<
headius>
are the other changes needed?
<
headius>
if we can just do that one change I'd feel better about 9.2.6 but I'll try to read those commits for your justifications
<
kares>
not sure - likely not. most of those were really minor - like (left-over) syncs on mutex methods
<
kares>
assumed it would be smt about interrupts ... how about COW breaking the handle_interrupted test?
<
headius>
well I'm going to try a different fix: iterate to copy those entries rather than addAll
<
headius>
minimalist change
<
headius>
that seems to run your example script just fine and still passes handler test
<
headius>
enebo: this may be the minimal fix for mutex issue
bga57 has quit [Ping timeout: 245 seconds]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
<
kares>
headius: that may fail due having null in the array returned from toArray?
<
kares>
if sourceStack changes betweek size and toArray call, right?
<
headius>
there shouldn't be any nulls in there but I'll add a guard
<
headius>
pretty sure the JDK collections reallocate if the size is wrong
<
kares>
nulls or if its too small to hold contents
<
headius>
but I removed that
<
kares>
than it should realloc, hopefully
<
headius>
it just does RubyHash h : stack
<
headius>
which will be a sync iterator but still an external iteration without deadlock
<
kares>
sounds better
<
headius>
I'll go with this and you can work on the cleanup and whatnot in your PR at your leisure
<
kares>
yeah that or a no-arg toArray()
<
kares>
hopefully I just forgot how the toArray(arg) works when array size != coll.size
<
kares>
... and thus am over paranoid :)
<
kares>
yeah, there might be nulls - it only does a new array alloc if it does not fill a.lenght < size
<
kares>
btw. I got a complained about the missing interrupt logic for Thread (all the tagged ones)
<
kares>
so there's more work to look forward next
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
<
headius>
kares: good to know about the array form...I hate that method anyway
<
headius>
normally I'd do .stream.toArray(i -> new RubyHash[i]) but who needs that extra stream
<
headius>
kares: definitely needs more work
<
headius>
I hate supporting half-baked features
<
headius>
kares: fwiw it's not terrible complicated logic so if you want to dig into that go right ahead
<
headius>
the basic plumbing is there
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
subbu is now known as subbu|lunch
Aethenelle has joined #jruby
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
Aethenelle has quit [Ping timeout: 250 seconds]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
<
headius>
darn that test
enebo has left #jruby [#jruby]
Puffball has joined #jruby
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
<
headius>
or the related ssl issue
subbu|lunch is now known as subbu
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
WedgeSEM has joined #jruby
<
WedgeSEM>
Hoping somebody can help me...
<
WedgeSEM>
damn... nobody home... alright
WedgeSEM has left #jruby [#jruby]
WedgeSEM has joined #jruby
WedgeSEM has left #jruby [#jruby]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
lucasb has quit [Quit: Connection closed for inactivity]
jmalves has quit [Remote host closed the connection]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]