This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: [RFC] PR libfortran/34669
FX Coudert wrote:
>> For this bug report, the problem is that gfortran sends the output of
>> -fdump-parse-tree to the assembler when compiling with -pipe. In the
>> case of the bug report, this breaks mk-kinds-h.sh and other
>> libgfortran configuration scripts.
> I'd go for the -S option (fixing mk-kinds-h.sh), as I don't like using
> stderr too much. But maybe the right way of doing it would be to
> create a file suffixed with .parse-tree (ie, create a.f90.parse-tree
> when compiling a.f90), similarly to what is done for dumping
> middle-end trees?
I also like the -S option more; at least as interim measure. It is also
a few milliseconds faster as the assembler is not called.
I don't think that we have to worry about users using -pipe and
-fdump-parse-tree at the same time.
Thus the libgfortran_pipe.diff patch is OK with me.
Tobias