Cloning reproducible-builds environment

[Chris Lamb]

[Keeping you in CC; like not subscribed to this "old" Alioth list but let me know if you are ]

Hi Jeff,

> The 2nd build fails on the official builds, too:

Indeed, this appears to rule out a reprotest-specific issue and believe I can reproduce this test failure by exporting:

  TZ="/usr/share/zoneinfo/Etc/GMT-14"

In case it helps it is fairly common for timezone code to mistakenly
assume that the hour offset is always less than ~12, perhaps based on
some idea that positive and negative timezones cannot overlap each
other (cf. GMT+1300 used for New Zealand DST IIRC).

Does this help?

I'm missing a little bit of clarity though as gscan2pdf 2.6.2-1 in
unstable/amd64 still FTBFS, but you seem to have added code to this
version (or an older one) to skip precisely this failing test:

        # reprotest has a environment which confuses the timezone maths.
        # So skipping the test including the hour until I understand it.
      SKIP: {
            skip 'skipping the full datetime + timezone test until I understand it better',
              1;
            is_deeply \@gmt, [ 2018, 12, 31, 12, 0, 0 ], 'datetime - timezone';
        }

> If you could show me how to run the second build first, I'll have a
> fighting change of fixing the problem in my package.

Assuming by "run the second build first" you mean re-ordering the
builds, I'm not sure how this would/could help? Like the above query
about the version skip, I think I'm missing something obvious here.


Best wishes,

-- 
      ,''`.
     : :'  :     Chris Lamb
     `. `'`      lamby at debian.org 🍥 chris-lamb.co.uk
       `-