[rb-general] Sphinx output always reproducible?

[W. Martin Borgert]

Thanks for your quick and extensive answer, Chris!

On 2018-04-10 22:15, Chris Lamb wrote:
> Finally, we could patch Sphinx to force the ordering. I'm not sure how
> difficult this would be, but it might be difficult to get to the Abstract
> Syntax Tree to get the "original" ordering (preferable) and I would prefer
> not to sort the contents for some undefined æsthetic reason.

That was, I confess, my first thought...

Sets are not ordered, so the original order in the source code
cannot be very important. For documentation any order is
acceptable - it is arbitrary now anyway. We would not make it
worse! :~) So why not just use sorted()?

Changing Sphinx would also make all software with similar issues
reproducible, not only one.

Cheers