[rb-general] Not detecting timestamp in gzip headers of the aspell-* packages

[Chris Lamb]

Hi Lior,

> > Can you give a specific, concrete example of a package/version
> > combination
[..]
> As suggested on my initial mail, look at aspell-ro.

As we are constantly rebuilding the archive knowing the version
numbers (not just the source packages) is pretty much essential
for dissection/comparison.

Otherwise, it is very easy to get a very very confusing race condition
where folks are looking at newer and different results, hence my
request for concrete package/pairs. Again, apologies this was not
clearer in my previous requests.

So, for clarity, I am looking at src:spell-ro (3.3.9-1) at:

  https://tests.reproducible-builds.org/debian/rb-pkg/unstable/amd64/aspell-ro.html

.. which is currently being reported as reproducible in unstable,
whilst src:aspell-ar-large (1.2-0-3) at:

  https://tests.reproducible-builds.org/debian/rb-pkg/unstable/amd64/diffoscope-results/aspell-ar-large.html

.. is currently marked unreproducible in unstable.

> they all lack the needed flags for gzip

So, I can reproduce these results locally on my own machine so it is
not a curiosity within our testing framework but rather in your
packaging.

As you can imagine I don't have the bandwidth to investigate deeper
right now (!) but glancing at the build log it appears like gzip is
called twice in aspell-ro (building the package in the clean target
somehow..?) but only once in aspell-ar-large.

The debian/rules files are substantially different (ie. it is not
really the same code at all) which would probably be the next step to
investigate why these packages differ are causing curious results.


Regards,

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