I haven't heard of any Ruby libraries binding to the VTD-XML parser (which I hadn't heard of before, thanks for the tip!). Sounds like an interesting project to take on. Possibly the easiest way to use it might be to use the Java VTD-XML parser library from JRuby.
Yeah, looks like FFI would be your best bet. The C gdbm extension is mostly just wiring the underlying APIs into Ruby, and it's only 1300 lines of C code :). Sounds like a nice side project. If you decide to port it, let us know so we can consider including it in JRuby core (since gdbm is usually distributed with Ruby).
No, we haven't done anything new with fork. We still have the experimental support by setting -J-Djruby.fork.enabled=true, but I've only seen it work in simple cases.
While performance was not a major theme for this release, there's a smorgasbord of lower-level things that may help certain applications. Things like numeric operations in the interpreter, some improvements to Fixnum math operations, and reducing the bytecode limit for JIT'ted methods so that we won't attempt to compile methods that are too big, thus wasting less time during compilation.
Thanks for the suggestion for a comparison between 1.3 and 1.4. We'll try to have some numbers to show closer to release time.
You're right, patch level 0 doesn't say much about the level of JRuby's 1.9 support. This is complicated by 2 reasons: 1) JRuby's own support is still missing a couple major features as mentioned; 2) 1.9 itself isn't fully baked.
Regarding 2), the RubySpec project has been targeting 1.9.2, and JRuby follows that as a result. Since 1.9.2 is still in preview releases, there is no publicly-known patch level to use anyway. If you think we should display something other than "0", please let us know.
![[Avatar]](/images/avatar26.png){kind=avatar}
Ruckus