Trouble installing libgcj on i686-pc-linux-gnu (RedHat 6.0)

Sandro Hawke sandro@roads.org
Thu May 6 11:47:00 GMT 1999


Tom writes:
> >>>>> "Sandro" == Sandro Hawke <sandro@roads.org> writes:
> 
> Sandro> I installed RedHat 6.0 (glibc-2.1.1) and today's (May 3) CVS
> Sandro> snapshots of egcs and libcgj.  EGCS built okay.  The libgcj
> Sandro> died on dtoa.lo, so I made the fix to mprec.h mentioned a days
> Sandro> ago.  Then the build continued on to natObject.lo, where I got
> Sandro> this:
> 
> Did you ever figure this out?
> 
> Sandro> /usr/local/build/libgcj-snapshot-1999-05-03/i686-pc-linux-gnu/libjava
/../../../../src/libgcj-snapshot-1999-05-03/libjava/java/lang/natClass.cc:88: m
ultiple definition of `java::lang::Class virtual table'
> Sandro> java/lang/Class.lo(.data+0x7d0):/usr/local/build/libgcj-snapshot-1999
-05-03/i686-pc-linux-gnu/libjava/../../../../src/libgcj-snapshot-1999-05-03/lib
java/java/lang/Class.java: first defined here
> 
> I seem to recall that this can happen if you accidentally build
> Class.h using gcjh.  Did this happen?  You can tell if Class.h appears
> in the build (not source) tree.

Class.h occurs in my source tree and not in my build tree.

> I also seem to recall that there might have been a bug along these
> lines introduced into egcs.  You'd have to check the egcs bug archives
> to be sure.

I'm not sure I understand.  I don't know what kind of bug in egcs this
might be.

> I'd rather work it out now.  It works every day for me, and I don't
> have a clue how I would reproduce your problems.

With glibc 2.1.1 ?  That's the most likely source of difficulty.
(Given that everything else seems to need to be patched to work with
it...  RedHat 5.2 to 6.0 is the least compatible move I've made since
SunOS4.0!)

Sorry, I don't really grok how egcs and the parts of libgcj all fit
together (what's a .lo file?) so I'm not sure where to poke.

    -- sandro


More information about the Java mailing list