method dispatch on Java objects / classes should try to find a matching varargs method

There is supposed to be a library released soon by one of the Rhino developers that implements Java-standard overloaded method dispatch at runtime. I believe that is our best bet for resolving all remaining method dispatch issues, since we can feed it direct Java types and coerced types until an appropriate match results.

Varargs is actually a pretty good idea, since again it's just a compiler trick in Java 5. We could easily consider methods that have an array as their last arg to be var-argable, provided all extra args in a call can map successfully to that array type.

I would love to see this in 1.0 timeframe, but it could get pushed past 1.0 and certainly won't be ready for 0.9.9.

Note: a patch that makes this varargs behavior work before 1.0 beta/RC release would very likely be accepted. Otherwise it may not be a priority until after 1.0.

I thought I saw some varargs stuff in Bill's patch. Either way, I think his patch will make this easier to fix.

This would be nice-to-have for 1.0, but it's not a high priority at the moment. Bumping to post RC for now.

Punting to post 1.0. Nice-to-have, but it would require too much code and would represent too much "new" or "changed" functionality to get into 1.0 this late in the game.

I think Bill's stuff could fix this in 1.1 timeframe.

Note, this bug should also fix the cases where you want to call a varargs methods without passing any of the varargs, as caused problems in JRUBY-1233.

Removing target release from issues that fit any of the following criteria:

• No known way to fix them
• Java integration enhancements out of scope for 1.1 release
• Other out of scope issues for 1.1

Ping. Just wanted to voice my vote in favor of this bug. It leaves me having to add syntactic noise in an otherwise beautiful language.

Some bit of googling revealed that rhino supports varargs: https://bugzilla.mozilla.org/show_bug.cgi?id=382457

Funny, I was just thinking about this issue this morning. Will take a look.

Bill: Give any more thought to varargs support? We could certainly use extra hands on JI stuff

I have finally implemented varargs support in af31aeb!

~/projects/jruby ➔ jruby -rjava -e "java.lang.System.out.format('foo %d, %d, %d, %d, %d', 1, 2, 3, 4, 5)"
foo 1, 2, 3, 4, 5

It's a little hackier than I like, but I'll mark this bug for 1.5 since it's working and has specs. More cleanup to come.

Marking as fixed. I did a pass to unify some of the method-construction logic, and additional cleanup will come with a future rewrite of the method-selection logic.