Re: gfortran 8?
Testing, and fixing the FTBFS bugs. For most of gcc / g++ we can just
fix the bugs "in-situ" in preparation for the changeover to default.
For Fortran, the changes in the mod file format means that packages
that need the new mod files wont build, and so can't be further tested.
e.g. metview is a mix of Fortran and C++, and regularly breaks on g++
changes (its got a bunch of old C++ code in it, not very standards
compliant). But it FTBFS because it needs grib_api.mod from eccodes, so
I need to force gfortran-8 for eccodes and metview to test.
Also note: there's an ICE in gfortran-8 for open-coarrays #897823. This
has a fix in the repo:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85507 ; might this make the
release for buster ?
Without it, we don't have coarrays for gfortran, one of its most
interesting features.
regards
Alastair
On 09/05/2018 11:33, Matthias Klose wrote:
> Hi,
>
> is there a reason to go ahead with the gfortran-8 change, before we make GCC 8
> the default?
>
> Thanks, Matthias
>
--
Alastair McKinstry, <alastair@sceal.ie>, <mckinstry@debian.org>, https://diaspora.sceal.ie/u/amckinstry
Misentropy: doubting that the Universe is becoming more disordered.
Reply to: