bug#36747: Official MesCC bootstrap binaries differ from my locally built ones

[Mark H Weaver]

Marius Bakke <address@hidden> writes:

> Mark H Weaver <address@hidden> writes:
>
>> I rebased 'wip-binaries' on top of current master (which includes the
>> recent 'staging' merge), and excluding the update of mescc-tools to the
>> git checkout.
>>
>> I built the bootstrap-tarballs for i686-linux and got the same hashes
>> that we've previously agreed on here.  I used "guix download" to load
>> the new bootstrap binaries into my store, and am now testing the
>> attached draft patch to 'core-updates'.
>
> Excellent, thank you!  The patches LGTM.
>
> I wonder if we should run these through a 'core-updates-next' branch to
> give ourselves a little time to bootstrap the different architectures.
>
> (also, it would be neat to get SQLite 3.29.0 in..)
>
> Thoughts?  I don't have a strong opinion, so do what you think is best.

I think we should continue to treat 'core-updates' as frozen.  These
slight changes to the bootstrap binaries to make them build
deterministically should almost certainly make no difference to anything
else in 'core-updates', so the only time we'll lose is the time needed
for Berlin to rebuild.

If we make any additional changes to 'core-updates', such as updating
SQLite or adding more architectures, it will likely cause additional
problems that need to be debugged.  This 'core-updates' cycle has
already taken too long, IMO.

Any other opinions?

      Thanks,
        Mark