<chrisseaton>
enebo: were you working on using symbols for internal identifiers?
<headius>
chrisseaton: he's probably off for the night but yeah
<headius>
long time coming to deal with identifier encoding issues
<chrisseaton>
I just don't want to transcode to Java strings and back all the time
<chrisseaton>
And hopefully share more byte[]
<headius>
are your method tables etc keyed on String?
<headius>
most of ours are but our symbol table keys off string and bytelist
<headius>
unless we need to deliver a symbol for reflective purposes we just use j.l.string internally
<chrisseaton>
Yes all identifiers internally are String
<chrisseaton>
What do you do with identifiers apart from lookup and deliver for reflection though? Who's using it as a j.l.String?
<chrisseaton>
I just feel like we're using it as a default but not for any good actual reason.
<headius>
JRuby started out with java string before I got involved...has not changed for two reasons I guess...
<headius>
1. inertia
<headius>
2. we integrate both ways with Java so there will always be a strong use case for lookups with java string
<chrisseaton>
I feel like it's when I realised for TruffleRuby that there's not really any reason to have a `TruffleRubyArray` etc class - no need for us to pretend we're Java so much
<headius>
right...we definitely are a JVM language and that means easy two way integration
<chrisseaton>
Yes you care more about Java interop at the moment than we do
<headius>
whatever enebo comes up with there will almost certainly remain java string paths to looking stuff up
<headius>
have you noticed going back and forth is a problem?
<headius>
the biggest reason we're doing it is because we get bug reports about identifiers not having the right encoding
<chrisseaton>
We think that's a problem but not ever seen it in practice
<chrisseaton>
I don't think there are any failing specs
<headius>
I'm not sure we fail any specs either, but there are MRI tests
<chrisseaton>
It's more of a case of 'wait why are we doing this again?' and it turns out no reason at all
<headius>
MRI mostly adds tests when someone reports a problem, so we are trying to iron out remaining cases
<headius>
and it will save us an extra indirection when we need to produce the symbol
pcarlisle has quit [Ping timeout: 240 seconds]
olle has joined #jruby
claudiuinberlin has joined #jruby
drbobbeaty has joined #jruby
shellac has joined #jruby
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]