[Debichem-devel] Bug#811449: Bug#811449: bagel: FTBFS: size_t is assumed to be the same size as long long

[Adrian Bunk]

On Tue, Jan 19, 2016 at 09:04:34AM +0100, Michael Banck wrote:
> severity 811449 important
> thanks
> 
> Hi Aaron,
> 
> On Mon, Jan 18, 2016 at 08:59:30PM -0500, Aaron M. Ucko wrote:
> > Builds of bagel for 32-bit architectures such as i386 have been failing:
> > 
> >   mpi_interface.cc: In member function 'int bagel::MPI_Interface::iallreduce(size_t*, size_t)':
> >   mpi_interface.cc:181:3: error: static assertion failed: size_t is assumed to be the same size as long long
> >      static_assert(sizeof(size_t) == sizeof(long long), "size_t is assumed to be the same size as long long");
> >   ...
> > 
> > Could you please take a look?
> 
> Well I can, but the fact that this is an assert() makes me think it is
> intentionally not supporting 32bit architectures.  Upstream has been
> rather opinionated that this code is geared at high-end supercomputers
> only.
> 
> We might have to restrict the architecture list to 64bit, I doubt just
> removing the above assert() will be the end of the story.

IMHO that became moot with the problems getting bagel compile on 64bit (sic)
with gcc 8 without exhausting the available RAM+swap on the buildds.

Please change the architecture list to
Architecture: alpha amd64 arm64 ia64 kfreebsd-amd64 mips64el ppc64 ppc64el riscv64 s390x sparc64

> Michael

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed