12:31 <byteit101[m]> master jruby with irb+readline changes is unusable for java objects for me

12:36 <byteit101[m]> https://github.com/jruby/jruby/issues/7323 and friends

15:00 <headius> Interesting

15:01 <headius> The right thing is probably putting back methods, maybe a subset of what you did, and those just will become packages you have to access with a string

15:01 <headius> I doubt we'll run into problems

15:02 <headius> And then it will turn out there's one guy out there who's written a Java wrapper for JRuby that uses a bunch of core methods as package names

15:03 <headius> But I imagine just a few reflection methods added back would make this work

15:03 <headius> We are a little bit aggressive in there right now

15:14 <byteit101[m]> I did nothing, the filer of that jira (vs my irb one) added back methods

15:14 <byteit101[m]> I have found the place we remove methods though

15:22 <headius> Oops ok

15:23 <headius> Yeah we should add back a subset

15:23 <headius> 9.4 is as good a time as any

15:31 <byteit101[m]> also this build of master isn't giving me foo= so I'll need to investigat

16:30 <byteit101[m]> headius: for packages, irb is erroring when searching ObjectSpace. That's bad in JRuby, yes?

16:30 <byteit101[m]> ObjectSpace.each_object(Module){|m|

16:30 <byteit101[m]> could we just remove the packages out of the modules OS?

18:04 <byteit101[m]> headius: new PR to add tests and fix some cases of setFoo not generating foo=

18:21 <byteit101[m]> How are objects added to the objectspace?

18:43 <headius> Yes I suppose that is also an option, we could add code into object space to skip those modules

18:44 <headius> We walk down the class hierarchy from Object for each_object(Module)

18:51 <byteit101[m]> I ask as the ctor argument is still adding the packages to the space

18:52 <headius> Yeah that's for a general object walking which is normally not enabled

18:52 <headius> We can support modules all the time by walking each class's table of subclasses

18:53 <headius> I'm trying to remember if there's a reason why we don't just filter them

18:53 <headius> We have other issues about this, probably worth digging them up to see what solutions we have proposed and shot down

20:59 <byteit101[m]> what mvn command should I be running for incremental testing? I often have to do a full clean and package because of generated classes

21:01 <headius> It used to be easier to just rebuild JRuby but I think the command is now something like mvn -pl core -pl shaded

21:01 <headius> We really need a rework of our build system so that incremental builds only rebuild the jar

21:11 <byteit101[m]> ah that seems to be it

21:11 <byteit101[m]> thanks!

21:23 <byteit101[m]> headius: I think I correctly removed the modules from OS in my new PR. I would like a review of that though. It does fix IRB today

21:24 <byteit101[m]> I have to fix PTY.spawn. I'm going to look at packaging libfixposix as mentioned in that issue this weekend

21:26 <headius> I'll have a look at your PR

21:27 <headius> Wrapping that library would be great! I've never gotten around to it unfortunately

21:27 <headius> It would make our process spawning much more feature complete

21:31 <byteit101[m]> I've been making so many PRs the past few days

21:32 <byteit101[m]> But yes, I ran into https://github.com/pry/pry/issues/1992 again and the best and most portable fix would be PTY.spawn

21:46 <headius> It could be something we roll into JNR and build it into the JFFI stub

21:46 <headius> Assuming it has a compatible license

22:09 <byteit101[m]> I'd like it for MRI too, so I'm probably going to package it separately at least, but that can be good too

22:09 <byteit101[m]> mostly because I'd like to add a pty.spawn_advanced or such that accepts fds

22:09 <byteit101[m]> naming is hard

22:18 <byteit101[m]> Options: IO.for_fd gains a jruby-internal kw argument, or I add IO.for_fd_jruby() as a new method ?

22:19 <headius> What's the kw?

22:19 <headius> Generally better to keep our changes separate so maybe even under JRuby::IO or something

22:20 <byteit101[m]> trying to find a nice way to pass the name and type of the pts into IO & file

22:20 <byteit101[m]> MRI:

22:21 <byteit101[m]> [#<IO:masterpty:/dev/pts/42>, #<File:/dev/pts/42>]

22:21 <byteit101[m]> JRuby: [#<IO:fd 18>, #<File:nil>]

22:21 <byteit101[m]> for PTY.open

22:21 <byteit101[m]> pty.rb calls IO.for_fd and File.for_fd in our code

22:22 <byteit101[m]> so we control both sides of the initialization

22:29 <byteit101[m]> or hmm...

22:36 <byteit101[m]> new PR for that

22:36 <byteit101[m]> so productive today :-)