Richard M. Stallman wrote:
The consequence is, that Sun
released OOo builds are obviously compiled with the same Java (and C++)
compiler as SO builds. Frankly we (RE) don't really care which Java
compiler we use but we certainly do care that we have to use only one.
I don't think that is a problem. It is important for OpenOffice.org
to be *tested* regularly with GJC (and GNU Classpath), so as to catch
problems quickly, but which compiler you use to build the release
binaries is a separate question.
I'm pretty sure that once GJC support is fully working there will be
regular builds with GJC even if we at Sun don't do these builds. Most
OpenOffice.org milestones (we do about 1-2 milestones per week) are
build many times outside Sun, and some will choose gcj and GNU Classpath
as Java support. I doubt that any breakage will go unnoticed longer than
a few days. Probably there will not be that much breakage at all once
everything is in place. If it doesn't work we have to think about
something else.
I think the easiest way to enforce the Java policy is to use the 1.3.1
API documentation for coding and asking the GNU classpath guys if in
doubt, as long they do not have the complete API in place.
That sounds like it should work. But rather than asking, you could
simply look at the list of features still missing; Mark Wielaard can
send it to you.
What RE will do is to provide the SUN developers with gcc-4.0.x and a
way to choose the gcj as alternative Java environment.
I am not completely sure what that means.
Oh, this just means that we will try to make it easier for Sun
developers to test changes to OOo with the gcj compiler.
Heiner
--
Jens-Heiner Rechtien
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
