gcj + xerces/xalan/tomcat/rhino/etc

[Anthony Green <green at redhat dot com>]

Hello gcj hackers -

    Building big packages of java code with gcj can be tricky
business.  Many projects are moving to tools like Ant, which don't
really work nicely with gcj today.  Perhaps one day... but until
then... I've created a cvs tree containing a number of open source
java packages which all build with gcj using the
autoconf/automake/libtool trinity we know and love.  It turns out that
once the basic infrastructure is in place, getting these things to
build is easy.  The trick nailing down the last few gcj/libgcj bugs
needed to get them to work!

    The tree contains imports of the latest stable xerces and xalan
releases, as well as a number of regexp packages, and miscellaneous
packages used by these things (cup, BCEL, ...).  I have more which
build and I'll commit over the next couple of days (tomcat, rhino,
tritonus, postgresql jdbc drivers).

Caveats:

    * You will need tools from the very tip of the GCC trunk in order
      to build and run these things.

    * Everything should build, but not everything works perfectly yet.
      xerces looks ok.  I'm having some problems with xsltc from xalan
      right now.  I could use some help.

    * You'll need lots of disk.  I did straight imports of the
      upstream source distributions.  Some of them could be trimmed
      (by removing some .jar files, .class files, etc), but there are
      ~2k java source files.  The libraries aren't small either.  I've
      appended a list to the end of this note.

      As I mentioned, I could use some help getting these into shape.
There's a really lame web page at http://sources.redhat.com/rhug.  It
should explain how to check-out the code and subscribe to the mailing
list.  

      I believe this is an important project for the following
reasons..

	  a) it will help nail down some of the few remaining (!) bugs
in gcj and libgcj that may only be discovered by building and running
large real world applications.

          b) it will eventually help popularize and educate people
about the use of gcj.

	  c) it will result in high-performance versions of important
modern tools and applications with no dependency on the use of
non-Free tools and execution environments.

If you agree with any of this --- please help!  Check out the code.
Build it.  Try to use it.  Fix it.  I'm happy to give out cvs write
access to people willing to help.  Thanks!

AG



On IA-32 Linux...

   1619018 Aug 29 14:56 lib-de-fub-bytecode.so.0.0.0
    162245 Aug 29 14:59 lib-gnu-regexp.so.0.0.0
     47994 Aug 30 06:26 lib-java_cup-runtime.so.0.0.0
     51804 Aug 29 14:52 lib-javax-xml-parsers.so.0.0.0
     95417 Aug 30 07:02 lib-javax-xml-transform.so.0.0.0
    479710 Aug 29 14:52 lib-org-apache-html.so.0.0.0
    586459 Aug 29 14:53 lib-org-apache-log4j.so.0.0.0
    396828 Aug 29 14:53 lib-org-apache-oro.so.0.0.0
    105833 Aug 29 14:53 lib-org-apache-regexp.so.0.0.0
    298125 Aug 29 14:52 lib-org-apache-wml.so.0.0.0
   2216135 Aug 30 07:02 lib-org-apache-xalan-xsltc.so.0.0.0
   1979286 Aug 30 07:02 lib-org-apache-xalan.so.0.0.0
   3452192 Aug 29 14:53 lib-org-apache-xerces.so.0.0.0
    211147 Aug 29 14:52 lib-org-apache-xml-serialize.so.0.0.0
   1099473 Aug 30 07:02 lib-org-apache-xml.so.0.0.0
   1016966 Aug 30 07:02 lib-org-apache-xpath.so.0.0.0
     14673 Aug 29 14:52 lib-org-w3c-dom-events.so.0.0.0
     80829 Aug 29 14:52 lib-org-w3c-dom-html.so.0.0.0