[Python-Dev] PyPI comments and ratings, *really*?

[Xavier Morel]

On 13 Nov 2009, at 00:34 , Jesse Noller wrote:
> That's because as an author/maintainer - we have methods of giving
> feedback and communication. Why not rate ( or auto-rate) packages on
> objective criteria?
> 
> E.g.: tests and test coverage, docs, installs on python version X, Y,
> Z, works on windows, etc?
Because there are lots of subjective criteria which are still very useful to users? The feeling of the API, the completeness of the library or its flexibility, etc…?

If pypi one day has a CPAN-style buildbot farm allowing it to test the package on any platform under the sun, that can be included, the tests can be included as well but given the number of testing solutions (and coverage discovery associated) that would be quite an undertaking.

And as far as docs go, what would be the criterion? "Has documentation"? How do you define "has documentation"? Has auto-extracted documentation from the docstrings? Has a README? Has a complete sphinx package? I don't think there's much that you can rate "objectively" about documentation.