Re: Two - Elementry Questions

[Alexandre Petit-Bianco <apbianco at cygnus dot com>]

Don Rozenberg writes:

> That is I would like to hope that gcj with machine-independent
> optimization and disabling run time checks it would be possible to
> realize a substantial performance enhancement.

Yes, provided that you build to native. It seems that you're trying to
compile to bytecode.

> My test case was to try it with the Apache xerces-1.4.2 package for
> XML processing.
> ...
> There were, of course, a few files that would not compile because of
> deficiencies in the library class files.

I didn't see that. Building xerces-1.4.2 to native with an up-to-date
gcj fails because of:

  http://gcc.gnu.org/ml/java-prs/2001-q3/msg00067.html
  http://gcc.gnu.org/ml/gcc-bugs/2001-08/msg00235.html

And something that I haven't seen until I tried to build xerces today
(it's been a long time I haven't but Xerces is returning to the set of
sources I routinely build:)

  ./org/apache/xerces/readers/UTF8Reader.java:171: Internal compiler
  error in optimize_mode_switching, at lcm.c:1256

I'll look into the Java PR soon. The other one is a gcc PR
(other/3970,) the last reported error needs to be investigated.

When compiling to bytecode, which seems to be what you're aiming at,
the only problem I see has been reported in the Java specific PR
mentioned earlier (java/4007.)

./A