Bug#962401: netcdf-fortran: please make the build reproducible

[Chris Lamb]

Hi Sebastiaan,

> Recording buildflags is a very common practice by upstreams

Speaking entirely anecdotally, after providing supplied patches for
hundreds of packages with respect to various reproducibility issues, I
think you would be surprised how few packages explicitly record
their build flags like this.

Regarding fixing it elsewhere in the toolchain, there are a great
number of subtleties and trade-offs involved. Solutions like "a dh
tool" (as well as many other solutions) are not workable in principle
or in practice. Like you, I wish it were otherwise.

This is not meant to be patronising in the slightest (I promise) and I
am not saying that a novel solution cannot be put forward at this
point, but I must gently point out that this build path issue is
something that the Reproducible Builds have been working on for the
past 5+ years. As you can imagine, we didn't look forward to or enjoy
the battles getting this into upstream GCC.

Anyway, the main point I want to make here is thank you for fixing
#962401 and for caring in general.

I don't think it would be productive to continue debating the merits
of a GCC flag (etc.) in a remote/leaf bug like this so I propose we
let this topic go quiet from here. Nevertheless, if you feel strongly,
please re-raise it on the following list where you can feel free to
quote me entirely:

  rb-general at lists.reproducible-builds.org



Regards,

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