This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: [patch, libgfortran] FIX PR19543 formatted output for logical(8) variables
- From: Tobias Schlüter <tobias dot schlueter at physik dot uni-muenchen dot de>
- To: Jerry DeLisle <jvdelisle at verizon dot net>
- Cc: Fortran List <fortran at gcc dot gnu dot org>, gcc-patches <gcc-patches at gcc dot gnu dot org>
- Date: Sat, 04 Feb 2006 19:54:17 +0100
- Subject: Re: [patch, libgfortran] FIX PR19543 formatted output for logical(8) variables
- References: <43E43F8A.2050107@verizon.net> <43E446FB.4080904@verizon.net>
Jerry DeLisle wrote:
> I still think that the types matching should be done though.
I'm not sure what it's meant to achieve. The ternary works the same for all
integer types.
> I realize that Tobi is working on the other part of this, was just curious
> looking at it. This case is weird. If I comment out the two test lines the
> prints come out correctly. If I allow the test lines, the prints come out wrong.
Well, I'm not working on it, and the last time I looked at it was more than a
year ago.
If I remember the specifics correctly, I determined that this was a middle-end
or back-end bug, it didn't seem to handle the boolean types set up by the
Fortran FE correctly. I don't know if anything has changed in that area in
the meantime.
> Heisenberg Uncertainty Principle. :)
Yes, that was my assessment at that time as well, if you look at comment #5
you will see that the results depend on how the backend decides to write the
constants to the assembly file, and this will change depending on the
constants in the surrounding code, as in your example.
- Tobi