jeremyevans has joined #jruby
jeremyevans has joined #jruby
jeremyevans has joined #jruby
_whitelogger_ has quit [Remote host closed the connection]
_whitelogger__ has quit [Remote host closed the connection]
ur5us has quit [Ping timeout: 244 seconds]
olleolleolle has quit [Ping timeout: 245 seconds]
olleolleolle has joined #jruby
olleolleolle has quit [Ping timeout: 265 seconds]
olleolleolle has joined #jruby
ur5us has joined #jruby
olleolle1lle has joined #jruby
olleolleolle has quit [Ping timeout: 256 seconds]
olleolle1lle has quit [Ping timeout: 256 seconds]
olleolleolle has joined #jruby
olleolleolle has quit [Ping timeout: 256 seconds]
olleolleolle has joined #jruby
jeremyevans has quit [*.net *.split]
ur5us has quit [Ping timeout: 264 seconds]
jeremyevans has joined #jruby
subbu is now known as subbu|away
drbobbeaty has quit [Ping timeout: 260 seconds]
drbobbeaty has joined #jruby
subbu|away is now known as subbu
marcheiligers[m] has quit [Quit: Idle for 30+ days]
subbu is now known as subb|lunch
subb|lunch is now known as subbu|lunch
jeremyevans has quit [*.net *.split]
subbu|lunch is now known as subbu
victori has quit [Ping timeout: 276 seconds]
<headius[m]> enebo: only things remaining for 9.2.17.0 are psych (which I will just punt to 9.3 because it isn't important) and the new .exe
<enebo[m]> ah ok I will make the new .exe
<headius[m]> I see nothing else recent that should go in
<headius[m]> byteit101: will be cycling back into 9.3 stuff so I have your PR up again
<byteit101[m]> awesome, let me know -if- *when* you have questions
victori has joined #jruby
<enebo[m]> pushed to jruby-9.2.
<headius[m]> neato
victori has quit [Client Quit]
victori has joined #jruby
victori has quit [Client Quit]
victori has joined #jruby
victori has quit [Remote host closed the connection]
victori has joined #jruby
<slonopotamus[m]> headius:
<slonopotamus[m]> whoops
<headius[m]> slonopotamus:
<slonopotamus[m]> Hi, I saw your PR about zlib, but didn't have a minute to test it yet) Going to try now.
<slonopotamus[m]> Is it OK that your fix exists in `jruby-9.2` branch but doesn't exist in `master` branch? I'm not sure about branching policy in JRuby repo.
<headius[m]> slonopotamus: we merge them peridically
<slonopotamus[m]> Okay. Then, I have to say that you actually fixed this thing!)
<slonopotamus[m]> * Okay. Then, I have to say that you actually fixed this thing!) Thanks!
<headius[m]> excellent!
<headius[m]> still an open issue for getting the rest of specs passing but at least this issue is resolved
<slonopotamus[m]> Hehe. It currently passes "deflates chunked data without errors" test but doesn't pass "deflates chunked data". Without looking into test details, this is at least funny.
<headius[m]> yeah the specs are not particularly good at isolating behaviors... I would guess the "without errors" specs just check that nothing blew up
<headius[m]> usually these specs don't get in because they don't test anything more than just testing proper outputs
<headius[m]> clearly if it errors it won't produce output
ur5us has joined #jruby
<headius[m]> don't want to scare you but I am going to look into 2.5.8 changes and do a stdlib update PR
<headius[m]> I don't think there will be much, and you can choose whether to include it or not based on diff
<enebo[m]> yeah I would like to see what changed in stdlib
<enebo[m]> It would be nice to be 2.5.8 since it probably looks strange we are not at .8
<headius[m]> that is the entire diff vrom 2.5.7 to 2.5.8
<headius[m]> ok the stdlib changes are basically nothing
jeremyevans has joined #jruby
<enebo[m]> yeah no stdlib at all from what I see
<enebo[m]> some very minor parser changes but even those are remove value_expr checks those could mean less void value expression errors but I suspect they are still caught elsewhere
<headius[m]> I will make sure bundled gems are updated
<enebo[m]> They can also mark unused variables but also I bet caught elsewhere
<enebo[m]> updated rake?
<headius[m]> the diff of actual stdlib files is super tiny so I will include this all in one PR
<headius[m]> checking... I think we may have updated some of these beyond 2.5.8 though
<headius[m]> because bugs
<headius[m]> oops
<headius[m]> they have several exts in default gems now that we do not
<headius[m]> super duper annoying but what are we suppose to do about e.g. etc
<headius[m]> bigdecimal
<headius[m]> ever feel like you're not wanted?
<headius[m]> I will also push a test update and see if anything new fails
<enebo[m]> lol
<enebo[m]> yeah so only the rake point is significant
<headius[m]> two local changes to stdlib did not make it back into the fork
<headius[m]> the more we can get away from using our fork the better
<enebo[m]> yep
<headius[m]> might be a short goal to get 9.3 off the fork and only copy files from CRuby 2.6 branch directly but I need to review the diff
<headius[m]> more stuff is in gems for 9.3
<headius[m]> enebo: I am thinking 9.2.17.0 should move to the io/console gem
<headius[m]> it is the same as what we had but now it includes missing features and passes all tests
<headius[m]> only in the gem version and master
<headius[m]> (which uses gem)
<enebo[m]> hmm
<enebo[m]> It is a risk but I am not sure how many people actually use it so I don't know how far missing features goes
<enebo[m]> My main question would be did anything basic change or is it really "the same"
<headius[m]> well let me get you a diff quick
<headius[m]> hmm diff is a little tricky because files we relocated to support both cruby and jruby in same gem
<enebo[m]> What things use it? irb.
<headius[m]> yeah
<headius[m]> some other command line things here and there to update text in place or set raw or password modes
<headius[m]> the added features are mostly cursor control and window size from termios
<enebo[m]> I guess part of me is not too into this unless someone reported us missing something.
<enebo[m]> I am hoping we can put out 9.3 soon
<headius[m]> ok this isn't bad
<headius[m]> the actual commit will simply remove our copies but this shows the diff
<headius[m]> some of these changes were crashers on other impls like `NCCS = 32`
<headius[m]> ours got lucky and did not segv
<headius[m]> FWIW I tested the FFI version against CRuby and it passes everything
<headius[m]> we can't pass everything because our pty.rb is broken and the tests use that to simulate a child tty
<headius[m]> I get it
<headius[m]> we were not including path and an io/console test expected it
<headius[m]> not the function name
<enebo[m]> but what is the difference between TCSADRAIN and TCSANOW
<headius[m]> one is what CRuby uses, the other is not 😀
<headius[m]> TCSANOW The change occurs immediately.
<headius[m]> TCSADRAIN The change occurs after all output written to fildes has been transmitted to the terminal. This value of optional_actions should be used when changing parameters that affect output.
<enebo[m]> but the first two lines was what it was and the next two is what it became but if this is just for MRI then fine
<headius[m]> so the intents was to flush but it did not quite do it the same way
<headius[m]> flush/apply immediately I mean
<enebo[m]> so it now will flush immediately ok
<headius[m]> if I change something, it either failed a test or didn't match the equivalent C code
<enebo[m]> much of it is just new stuff so I am not really against that
<enebo[m]> c_cc vs cc_c is something I don't get at all but I imagine that was a typo?
<headius[m]> and this adds support for the kwargs that `raw` and `getch` are supposed to support
<headius[m]> yeah typo... I did not see any reason it was named differently so I just fixed it
<enebo[m]> yeah kwargs change looks interesting in that it passes them down
<enebo[m]> well if the code here is obviously hit (for the changes) and we work on linux/mac/windows I guess I am game
<headius[m]> I will include in 2.5.8 PR then since that is the last stdlib piece that is not up to date with latest
<enebo[m]> I do really want to clamp down on what we fix there. We know we have plenty of bugs we will probably never fix at this point
<headius[m]> anything where we match 2.5.8 but there is something newer I do not update
<headius[m]> this one just stuck out for me because I know the old io/console we ship is broken for several features
<headius[m]> and the gem one is 100%
<enebo[m]> I almost exclusively would like it to only be raised focus on a user reported problem
<enebo[m]> but this is not a lot of changes and if it leads to problems we can just for .18
<headius[m]> yeah I can't say anyone has reported the missing features
<headius[m]> other than me while trying to get our code into the gem
<enebo[m]> So the other observation about these point releases which is maybe obvious but the more we have the more we get reported problems
<enebo[m]> That has nothing to do with quality
<headius[m]> yeah people see the activity and try newest sooner
<enebo[m]> It likely means more releases gets more people trying stuff
<enebo[m]> In that sense the notion of a .18 or .19 based on a couple of fixes tends to almost be marketing
<enebo[m]> but with that said I think 9.3 should be most energy as I think that will garner a lot more interest
<enebo[m]> Each time I see some image manip. Q for JRuby someone will reply..."but it has not been released for n months. I assumed it is dead"
<enebo[m]> image_voodoo just works. This notion that software has to be released or people have no interest is still a phenomena which always surprises me :)
<headius[m]> I agree and do not want to have to do .18 any time soon
<headius[m]> once this 17 is settled I will be back on trying to close 9.2 issues and PRs
<headius[m]> er 9.2
<headius[m]> AGH 9.3
<headius[m]> I dunno how to deal with that... software that is "done" looks "dead", but software that spins a release every week is probably super unstable
<headius[m]> or at least the risk is high
<headius[m]> enebo: this 2.5.8 update is ready to go then
<headius[m]> I will not do tests in this PR because there are a lot of places we have patched our tests and I don't feel like reviewing that for a minor update
<enebo[m]> ok I looked at the rest. So I say merge it assuming nothing is broken
<headius[m]> ok, after it is green I will merge
<headius[m]> enebo: there is one failure in io/console test but I think it is an out of date test
<enebo[m]> hmm yeah I guess if it is gone later
<headius[m]> I believe I have an explanation
<headius[m]> I think CRuby subprocess launches do not inherit tty the same way so this test sees a console when it shouldn't
<headius[m]> part of my fixes updated it to acquire the console like CRuby, which is likely why it started failing
<headius[m]> the only thing that tests it whether IO.console is nil if you don't have a tty
<headius[m]> I have pushed an updated test file and excludes (all excludes go away and test_noctty was added)
victori has quit [Quit: ZNC 1.8.2 - https://znc.in]
<headius[m]> ok, 9.2.17.0 is done... versions updated, stdlib updated, issues and PRs closed or rehomed
<headius[m]> also merged to master
<headius[m]> kalenp: I send something to release@jruby.org but not sure if it propagated... 9.2.17.0 next week
<kalenp[m]> Thanks for the heads up! We've upgraded to 9.2.16 without any issue so far.
<kalenp[m]> Though I suspect that the IO.copy_stream story isn't entirely done. I haven't dug into the details yet, but it's still misbehaving in our integration tests while passing in the unit tests.
<headius[m]> hmm
<headius[m]> that blasted thing
<kalenp[m]> hopefully next week I can dig into what's going on now. the hard part is that it doesn't show up in our unit tests, but only when we're actually copying to a stream produced by our Mail gem. so seems to depend on some details of the particluar IO object that's on the receiving end
<kalenp[m]> I thought I'd captured it all in that initial bug report, but apparently not
<kalenp[m]> gotta hop now. thanks again for the heads up on the next release
<headius[m]> hmm well let me know if you can isolate the problem
ur5us has quit [Ping timeout: 264 seconds]
victori has joined #jruby
victori has quit [Client Quit]
victori has joined #jruby
victori has quit [Client Quit]
victori has joined #jruby
ur5us has joined #jruby