Re: libtool is hiding errors from failed GCC build

[Alexandre Oliva]

On Nov  5, 2001, Bryce McKinlay <address@hidden> wrote:

> We have an i686 bootstrap failure at present in libjava which only
> occurs when a file is compiled with -fPIC. Confusingly, libtool is
> hiding this error because it sends output from the -fPIC compilation
> to /dev/null.

> Alexandre, is this patch ok to install in GCC?

I'm afraid not.  If we're to change libtool like this, we should save
the output of the first and the second runs, and only display it again
if they differ.  The problem is that letting the output go through to
the stdout while still redirecting it gets tricky if you don't have
tee.


Perhaps we should redirect the output of the second run to a file, and
print it only if compilation fails?  This would be much easier to
implement, and guarantee the property of only displaying the output
the second time when it differs from the first time (because otherwise
we wouldn't have gotten to the second run, as you wrote yourself).

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  address@hidden, redhat.com}
CS PhD student at IC-Unicamp        address@hidden, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me