[Debian-med-packaging] Bug#970875: Bug#970875: mash: inconsistent results on 32 vs 64 bits architectures

[Étienne Mollier]

Sascha Steinbiss, on 2021-02-20 15:41:44 +0100:
> Hi,
> 
> >>> Should we tag this 'upstream'?
> >>
> >> Ah, good finding!  Yes, I believe it would make sense to tag it
> >> upstream.
> > 
> > Probably.  However, what about providing it for 64bit architectures
> > only?  I mean the practical relevance for 32 bit architectures is
> > very limited and I consider it burned developer time if we care
> > for something where we know it provides different results than
> > expected and is not used anyway.
> 
> Yes, agreed. I would always be in favour of restricting archs and RM the
> rest -- but then again I don't have any stakes in anything non-amd64.
> 
> Any other opinions?

I don't know by which end to solve the 32 bits failures here, so
it might make sense to drop 32 bits.  None of the reverse
dependencies I saw were available on 32 bits release
architectures, either because of missing build dependencies
(ariba, plasmidid) or failure to build from source (kleborate).

As far as I can see, the working set would be:

	Architecture: amd64 arm64 mips64el ppc64el s390x alpha ppc64 riscv64

(I don't include m68k on which kleborate built successfully,
 because the test suite is not run on that architecture.
 Otherwise I suspect it would fail.)

Cheers,
-- 
Étienne Mollier <etienne.mollier at mailoo.org>
Fingerprint:  8f91 b227 c7d6 f2b1 948c  8236 793c f67e 8f0d 11da
Sent from /dev/pts/2, please excuse my verbosity.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/debian-med-packaging/attachments/20210220/d4d0efae/attachment.sig>