No need for data and the loop. Just use assert directly on the values.

This look unrelated to the change.

I know... But I noticed it was not tested... Maybe I can do a different PR.

Hmm, did you accidentally revert some commit? (bddd08977e45b0940b6a6f16d3c7ae7d14cbb346)

@lholecek about your last comment: yeah, I think I did some mistake. It should be fine now.

It should be ok now

Hmm... but shouldn't it not make an HTTP request in this case? Why do we assert that requests.request was called?

"determinate" is not a word, you want "determine" :-)

Also it's always a good idea to tell the user exactly how to provide the missing value, in errors like this. I would suggest to keep it simple, something like this:

Specify product version using --product-version

Maybe we should have two separate functions for guessing the product version, one which takes a compose id and one which takes a dist tag (or Koji build target I guess)... I am worried that .startswith('f') might be too broad if that logic is applied to compose ids, which it looks like it will be here.

Also, I wonder if there is similar logic in rpkg/fedpkg/rhpkg which we could reuse somehow... I think it knows how to guess dist tags and build targets based on git branches which is similar.

@dcallagh Thank you for your comments. I changed what you suggested.
I think it should be fine now.

For what concerns the fact that we can probably find a similar logic in rpkg/fedpkg/rhpkg: I checked a little and I didn't find what I needed (probably lack of knowledge of those software by my side). Maybe we can talk about it in our meeting tomorrow (Wednesday).

Here, 'rhel-7' should be 'epel-7' instead.

Corrected what Ralph said and it should be more readable now.
If you don't like it let me know and I'll change it again.

:+1:

+1 though this slows tests considerably (py.test-3 tests/test_cli.py::test_create_waiver_no_product_version takes 2s).

Also, I wonder if there is similar logic in rpkg/fedpkg/rhpkg which we could reuse somehow... I think it knows how to guess dist tags and build targets based on git branches which is similar.

Guessing build target from git branch is easy (at least for rpm builds): you just slap -candidate and the end (unless it's master branch in which case you just return rawhide).

Dist tags are more fun. Fedpkg does this. Rhpkg is significantly more complicated. Look for load_rpmdefines method in rhpkg/__init__.py. Either way the code is not really reusable.