GitLab

I see a red door and I want it painted black.

see: https://github.com/mgedmin/check-manifest/issues/115#issuecomment-625096256

those will be removed from scipy in version 2.0

> DeprecationWarning: scipy.empty is deprecated and will be removed in SciPy
> 2.0.0, use numpy.empty instead

This new function enables to compute a confidence from a distance. When
iterating over the aligned pairs, one can use this function to obtained a
confidence from the distance between the pair's elements.

For all aligners, the method `get_aligned_pairs` returns an iterator over
tuples of aligned pairs : (refitem, targetitem, distance). The PipelineAligner
was iterating over (refitem, targetitem) ; the distance was removed.

This patch fixes this, and make the content of the variable `pair` more explicit.

closes #1

This new parameter, `repeatable`, enables one to obtain the same results, with
the same parameters (documents, signature length, etc).

The MinHashing technique is based on randomly chosen hash functions. In order to
get the same results at each calls (when `repeatable` is set to True), we set
the random seed to a predetermined value (and customizable if need be). With the
same random seed, one should get the same hash functions, and therefore the same
documents signatures, and… the same results.

BREAKING CHANGE: this completely drops python2 support

this is the last version with some python2 support

heptapod has been upgraded to a newer version. Let's now simplify the gitlab-ci.

there is issues with the version 0.42 (the last release). Issues seems to be
related to the ignore folders, and the prune statements in the MANIFEST.in

related: https://github.com/mgedmin/check-manifest/issues/115

there is a issue with loading remote DTD. This issue is not resolved (and do we
want to keep this class `ValidXHTMLPrettyPrint` ? I’m not sure).

As this may me removed soon, I suggest to allow failure on this tests while we
decide whether we remove it in 1.0.0 or not.