Oleg Pykhalov <go.wigust@gmail.com> writes:

> Hello Marius,
>
> Apologies for the late reply and thank you for review!
>
> Marius Bakke <mbakke@fastmail.com> writes:
>
>> Oleg Pykhalov <go.wigust@gmail.com> writes:
>>
>>> * gnu/packages/web.scm (python-clf): New variable.
>>
>> [...]
>>
>>> +(define-public python-clf
>>> +  (package
>>> +    (name "python-clf")
>>> +    (version "0.5.7")
>>> +    (source
>>> +     (origin
>>> +       (method url-fetch)
>>> +       (uri (pypi-uri "clf" version))
>>> +       (sha256
>>> +        (base32
>>> +         "0zlkzqnpz7a4iavsq5vaz0nf5nr7qm5znpg1vlpz6rwnx6hikjdb"))))
>>> +    (build-system python-build-system)
>>> +    (propagated-inputs
>>> +     `(("python-docopt" ,python-docopt)
>>> +       ("python-pygments" ,python-pygments)
>>> +       ("python-requests" ,python-requests)))
>>> +    (home-page "https://github.com/ncrocfer/clf")
>>
>> I noticed in the '.travis.yml' that it has a test suite that can be run
>> with 'nosetests'.  Can you try overriding the check phase with one that
>> runs it?  That way we are likely to catch regressions early.
>
> Release tarball from pypi doesn't include test_clf.py.  So I pulled it
> from git master branch.  Almost all tests require internet connection.

Odd that this file would be taken out.  Anyway thanks for including it.
I wouldn't bother binding the package in a let for a variable that's
only used once as a native-input, but no strong opinion.  LGTM.