<enebo[m]>
I just fixed some warning but it must not be those
<enebo[m]>
the redefinition ones I have no idea on
<enebo[m]>
headius: do you know if this is new on HEAD or did we have it before?
<headius>
pretty new head, like a few days old, but I don't know if the warnings are new
<enebo[m]>
This appears to be marking a new scope and thinking something is not used
<headius>
I don't remember seeing them when I was generating apps for RubyConf
<headius>
or at least not as many of them
<enebo[m]>
the redfinition is something else
<enebo[m]>
not a parser thing
<enebo[m]>
I would be surprised if that is new
<headius>
I'm not sure if they affect anything but they are noisy
<enebo[m]>
or I don't recall changing anything that would cause that but I did fix an extremely noisy warning in the parser recently
<headius>
well keep it in the back of your mind... a lot of warnings just went away in 3.0/3.1 so we may have excess at this point, but some of these may be valid
<enebo[m]>
I will look into the date_time_parser.rb thing because it looks like something is not tracking use/def properly for variables
<enebo[m]>
I am laying a little money aliasing is broken in the concurrent-ruby warns
<enebo[m]>
e.g. we are warning for some case where we shouldn't
<enebo[m]>
lol. I am confused. I do see my fixing anything involving warnings on parser but I know I did for ...
<enebo[m]>
maybe I didn't merge that
<headius>
I have been a bit scattered with PRs lately too
<headius>
I put my call info PR on hold for now but landed several others last week
<enebo[m]>
ah yeah I guess I didn't even fully finish that PR
<enebo[m]>
that should not affect what you see but it is a massive warn tsunami during ar tests
<enebo[m]>
I have been trying to figure out some bad issue with File.absolute_path on windows. We have some serious problems to work through when we are non-ascii paths
<enebo[m]>
all those redefinitions might be valid. We are replacing java stuff over original defs in both cases
<headius>
that may be due to Windows having non-UTF-8 as standard encoding
<enebo[m]>
well it is a combination of things
<headius>
or filesystem encoding or something
<enebo[m]>
some methods just get Java default charset strings
<enebo[m]>
some methods make Ruby string from UTF-8 to filesystem encoding
<enebo[m]>
some methods will then try and use the combo of UTF-8 + CP-1252 to build a Java string
<headius>
yeah it's not handling the combination of encodings at the edges very well
<headius>
we don't see it because we default to utf-8 most of the time and other platforms are good with that
<enebo[m]>
I do think if something properly handles recombining Ruby strings between those two encodings it would maybe all start working (albeit we should be converting a lot more methods to use file system encoding)
<enebo[m]>
and most people are not using non-7bit paths
subbu has quit [Ping timeout: 252 seconds]
<enebo[m]>
I got super confused by walking through why we do not print out mbc cp-1252 to console (not related to the bug at hand) but in the end our use of console() in java should just be replaced by native call
<headius>
the work byteit101 is doing will mesh here as well since he's putting together native IO and process stuff on Windows... so that may help us improve this compat
<enebo[m]>
if all is ascii he may not notice but yeah
<headius>
byteit101: I know I have not been able to help on Windows... sorry about that
<headius>
crazy week for me
<headius>
huh, IBM cloud has Z series machines on my account now
subbu has joined #jruby
<byteit101[m]>
I think I'm mostly done except for that weird hanging issues with reading pipes. I'm also not sure I want to expand the scope of what I've done that much to other native IO stuff, as this work started as a way for me to get a working PTY.spawn on linux :-)
<byteit101[m]>
That's why I've been trying to reuse IO.pipe and such
<headius>
ok
<headius>
ugh why can't anything work first time... set up an IBM Cloud instance for running stuff for RailsConf... gave it my pubkey... refuses to log in
<headius>
impossible to search for
<headius>
ugh ok... the fedora coreos image is busted on IBM cloud for some reason
<headius>
spun one up with Centos and it worked first time
<headius>
I guess I'll be testing with CentOS for now then
<enebo[m]>
generated sources have empty blocks of code which end up with some implicitnilnodes vs null (which gets rid of need for null checks all over) but it saw a nil which is not being assigned to anything
<enebo[m]>
normal finished code would not normally have no code things to see this
<rcrews[m]>
I'm pretty handy with this kind of work
<rcrews[m]>
Otherwise, we can sync up in two weeks or so
<rcrews[m]>
Let me know if you have concerns, advice, or guidelines for me
<enebo[m]>
rcrews: cool
<enebo[m]>
headius: Those concurrent-ruby errors are whack
<enebo[m]>
err warnings
<enebo[m]>
They occur on the first line of actual code in a class: DEFAULT_MAX_POOL_SIZE = java.lang.Integer::MAX_VALUE # 2147483647
<enebo[m]>
Note there is no methods being redefined. So I am guessing refinements or prepend is making new method table and noticing an entry exists or something like thaT?
<enebo[m]>
Done for today but if you feel the need to remove our last warnings it is bound to be one of those two things