tjohnson has quit [Quit: Connection closed for inactivity]
pawnbox has joined #jruby
<GitHub92>
[jruby] atambo reopened issue #3094: NoMethodError being reported as "UndefinedValue.java:52:in `undefinedOperation': java.lang.RuntimeException: IR compiler/interpreter bug: org.jruby.ir.operands” in jruby-9.0.0.0.pre2 https://git.io/vtMyE
uzzaldevkota has joined #jruby
uzzaldevkota has quit [Client Quit]
yfeldblum has quit [Remote host closed the connection]
yfeldblum has joined #jruby
yfeldblum has quit [Ping timeout: 264 seconds]
bga57 has joined #jruby
yfeldblum has joined #jruby
lumrandir has joined #jruby
yfeldblum has quit [Read error: Connection reset by peer]
<GitHub23>
[jruby] kares commented on commit 82d8c47: woow - totally did not know this was added to Ruby ... what an idea to get constants feel even less constant :) https://git.io/vgtzj
<GitHub178>
[jruby] thomaswue commented on commit e81bd13: Behind a Truffle boundary, this directive has no effective. Is this case something that can ever happen anyway? https://git.io/vgqOy
<GitHub138>
[jruby] thomaswue commented on commit e81bd13: Can you add some unit tests for this flattening? https://git.io/vgqOH
skade has quit [Quit: Computer has gone to sleep.]
n00bdev has joined #jruby
lumrandir has quit [Quit: Leaving.]
n00bdev has quit [Ping timeout: 252 seconds]
bbrowning_away is now known as bbrowning
rashmirathi has quit [Remote host closed the connection]
skade has joined #jruby
nirvdrum has joined #jruby
tomjoro has quit [Remote host closed the connection]
hobodave has quit [Quit: Computer has gone to sleep.]
vtunka has quit [Quit: Leaving]
vtunka has joined #jruby
n00bdev has joined #jruby
n00bdev has quit [Ping timeout: 260 seconds]
mrmargolis has joined #jruby
mrmargolis has quit []
enebo has joined #jruby
<GitHub81>
[jruby] enebo commented on commit 82d8c47: @kares I felt the same way but 'A ||= 1' has always worked so I think they were just trying to be consistent. https://git.io/vgq72
<GitHub116>
jruby/ruby-2.3 0f21ba1 Thomas E. Enebo: Fixes #3639. superclass production in parser became less restrictive in 2.3
<GitHub8>
[jruby] enebo closed issue #3639: superclass production in parser became less restrictive in 2.3 https://git.io/vgmgn
tjohnson has joined #jruby
subbu|away is now known as subbu
skade has quit [Quit: Computer has gone to sleep.]
shellac has quit [Quit: Ex-Chat]
n00bdev has joined #jruby
ITXpander has quit [Ping timeout: 256 seconds]
n00bdev has quit [Ping timeout: 260 seconds]
<eregon>
enebo: Hi!
<enebo>
eregon: hello
<enebo>
eregon: so when we merge ruby-2.3 you sync up spec commits back to ruby/spec?
<eregon>
enebo: I looked at the hexadecimal floats, but it seems fairly hard to share logic with the existing parser, maybe it should just fork on a different path quickly (we should not allow '.', 'E', etc in that mode I guess)
<eregon>
that's one way, yep
<eregon>
we could also sync them to ruby/spec before that
<enebo>
eregon: without propagating a flag to say hex string through the parser yeah perhaps
<eregon>
yeah we would need to check the flag a lot
<enebo>
eregon: well we pass now in a branch which will merge and MRI always has passed but just never tested
<enebo>
eregon: yeah I have not looked at the parser in a long time but I believe based on how I make recursive descent parsers I push all state through the functions with no fields present
<enebo>
hey I should rewrite it as parser combinators!
<eregon>
yep, essentially it would just share the parsing of the first bunch of digits but then go a different path
<enebo>
eregon: so this will just lead to some duplicated methods essentially
<eregon>
yeah, writing it in antlr as chrisseaton does sounds tempting as well
<enebo>
eregon: well mostly duplicated
<eregon>
yeah, sounds fine enough
<enebo>
I am trying to remember does antlr still require runtime lib?
<chrisseaton>
it does but it's 32 KB I think
<eregon>
and since Java's Double.valueOf can already parse this stuff, it just a matter of removing '_' I think
<enebo>
that was the biggest issue with it besides being slow
<chrisseaton>
Apparently Antlr 4 is much faster, but I haven't tried for myself yet
<enebo>
chrisseaton: yeah I was told that about Antlr 3.x too
<enebo>
:)
<chrisseaton>
my main worry is that the community is not very active
<enebo>
chrisseaton: personally I think the features of nice error reporting have a non-zero cost
<chrisseaton>
the IRC room is dead, and the documentation is crap
<enebo>
I think their AST generation stuff is cool too but there was something non-optimal about that
<enebo>
anyways it is a cool project
<enebo>
I just wish they could generate fully and not require runtime lib
<enebo>
I think that would make it a better candidate
<enebo>
eregon: I think we used to do that but there was something else like non-strict vs strict
<enebo>
eregon: something Java just did not handle the same
<eregon>
so well, Double.valueOf(rubyString.replace("_", "")) might be a quick fix in practice when forking to hexa mode :)
<enebo>
oh hahaha yeah perhaps so
<eregon>
yeah there are already plenty of flags there
<enebo>
eregon: fwiw it has been in Ruby forever and never reported
<enebo>
like pre Ruby 1.8
<eregon>
well, yeah this kind of inherited-from-C stuff tends to be known by very few
<enebo>
so any impl probably is ok since no one uses the feature :)
<eregon>
indeed, but the specs are not too kind either :D
<enebo>
yeah the specs make all bugs the same importance
<enebo>
Like I probably wasted my time today fixing: ‘class A 1; end’ as parsing
<enebo>
but it fixed a MRI test :)
<eregon>
we are always free to prioritize what's most important first though
<enebo>
eregon: yeah which is why I did not run off and start working on 0x1p2
<enebo>
eregon: but I guess I am guilty in suggesting you fix it :)
<eregon>
ahah, well I gave it a quick try and re-prioritized
<enebo>
I am just happy when I fix something and it also fixes it in truffle…I feel like I did twice as much work in half the time :)
camlow32_ has joined #jruby
<eregon>
yeah, same for me :)
camlow325 has quit [Ping timeout: 256 seconds]
skade has joined #jruby
bbrowning is now known as bbrowning_away
skade has quit [Quit: Computer has gone to sleep.]