search

FCS-analysis / multipletau

python multiple-tau correlation algorithm
https://multipletau.readthedocs.io
BSD 3-Clause "New" or "Revised" License
19 stars 9 forks source link

Please make the version (and documention build) reproducible. #16

Closed lamby closed 6 years ago

lamby commented 6 years ago 
Whilst working on the Reproducible Builds effort [0], we noticed
that python-multipletau could not be built reproducibly.

This is because it uses the current date which gets embedded into
the documentation file if you are building from a tarball release
and not Git.

Patch attached that uses SOURCE_DATE_EPOCH [1]. This was originally
filed in Debian as #912957 [2].

 [0] https://reproducible-builds.org/
 [1] https://reproducible-builds.org/specs/source-date-epoch/
 [2] https://bugs.debian.org/912957
paulmueller commented 6 years ago

Sorry that my manual versioning approach does not work out for the reproducible builds effort. I cannot accept the current PR without modifications, because it would break my versioning approach.

How are you retrieving the source tarball of multipletau? If you obtained it from PyPI (recommended), then the versioning should be correct (there is a _version_save.py) file in that tarball.

If you are retrieving the source tarball from GitHub (not recommended), then the date (https://github.com/FCS-analysis/multipletau/blob/master/multipletau/_version.py#L119-L125) is used, which is what I believe is the problem here.

If you want to be as close as possible to upstream and have to use the GitHub repo, then I would accept this PR if the changes were incorporated in #3 instead of in #1.5. I could do that within the next few days if you are too busy.

lamby commented 6 years ago

because it would break my versioning approach.

How so? Only if SOURCE_DATE_EPOCH is present, no?

How are you retrieving the source tarball of multipletau?

I'm not the maintainer of this package in Debian so I cannot canonically say...

paulmueller commented 6 years ago

It turns out that this might just be a bug and I fixed it in https://github.com/FCS-analysis/multipletau/commit/d2734df4680e615c3577b22d88670d972956da88.

@mestia Please try again with multipletau ~0.3.1~ 0.3.2

lamby commented 6 years ago

(Did you mean mestia or myself?)

mestia commented 6 years ago

(Did you mean mestia or myself?)

I guess he meant me, I am preparing the new upload which hopefully will solve the problem.

lamby commented 6 years ago

Yay!

mestia commented 6 years ago

Sun is shining for python-multipletau at https://tests.reproducible-builds.org/debian/rb-pkg/unstable/amd64/python-multipletau.html I think the issue can be closed now. Thanks Chris for reporting and Paul for the quick fix!

paulmueller commented 6 years ago

Thank you!