CFP: USENIX Java Virtual Machine Symposium

Matt Welsh mdw@EECS.Berkeley.EDU
Thu Oct 12 10:13:00 GMT 2000


I would really like to encourage the GCJ team to put together a paper
for this symposium. We're willing to grant an extension to the submission
deadline if you need it -- a paper on GCJ is exactly the kind of thing
we're looking for.

The whole nature of GCJ -- its design, performance, and challenges -- is
certainly of interest for this conference. Given that Java is such a
"dynamic" language (dynamic class loading, dynamic compilation,
optimization, etc.) it's interesting to see how GCJ has managed to
develop a "static" implementation. The runtime classloading mechanism is 
worth talking about. Also, what optimizations does GCJ perform that a 
JIT compiler has trouble doing well? (For example - method inlining, etc.) 
What performance advantages would you expect GCJ to have over "the best
possible" JIT (not including dynamic optimizations)?

More focused aspects of GCJ - such as the AWT implementation, challenges
in implementing Sun's class libraries, and debugging - are interesting
as well. Is GCJ being used on any non-PC platforms (such as very large
servers or embedded systems)?

Matt Welsh


More information about the Java mailing list