Re: [PATCH] Java: Standard comformant arithmetic

[Andrew Haley <aph at pasanda dot cygnus dot co dot uk>]

> Date: 22 Jul 1999 08:39:51 -0000
> From: Andrew Haley <aph@cygnus.co.uk>
> 
> > From: Tom Tromey <tromey@cygnus.com>
> >
> > Alexandre> % egcj -print-prog-name=jc1 
> > Alexandre> gcj: libgcj.spec: No such file or directory
> > 
> > Alexandre> Err, does it really need libgjc.spec to tell me where jc1
> > Alexandre> lives?
> > 
> > I think this is just a bug in the new code.  I think it should only
> > look for the spec file when actually trying to compile something.
> 
> Right, that's a bug.  We need to add more logic to determine if phases
> which need the specfile will be run.  I'll have a look at this.

I have been looking at this and it's quite hard to fix.  gcc delays
the processing of the "-print-*" commands until all the specfile and
path processing has been done, presumably because until then location
of the subprogram hasn't been determined.

We could suppress the reading of libgcj.spec if any of the "-print-*"
commands are used.

We could perhaps print a warning if libgcj.spec wasn't found, rather
than treating such a failure it as a fatal error.  However, that would
probably require a change to code in the top-level gcc.

Andrew.