On 08.08.2016 16:08, Floris Bruynooghe wrote:
On 8 August 2016 at 06:48, Ronny Pfannschmidt
<[email protected]> wrote:
That would break external code
Uhm, so we've committed to mainting py._code in two places? Makes me
wonder what the motivation to merge into pytest was.
no, pylib will no longer evolve, its a relative dead end
the main reason to copy it is, that now that it is mostly an internal
api, we are allowed to change it without breaking downstreams - we moved
it so we can change it again.
Pylib is maintenance-only
While it's not terribly actively developed there's still scope for new
features I think. E.g. the PR for PEP 519 support counts as a new
features, I wouldn't want to block it.
the ideal case there is to eventually get rid pf py.path - in particular
since pathlib/pathlib2 is the way to go on more recent python and
py.path is really hairy to evolve
pep 519 as addition is not really a evolution i'd argue - however we are
unable to ever do that correct with py.path due to its broken bytes vs
unicode semantics that we cannot fix without subtle breakages of
downstream (hence the idea of eventually no longer needing pylib as its
under bad constraints wrt evolving it)
Floris
-- Ronny
_______________________________________________
pytest-dev mailing list
[email protected]
https://mail.python.org/mailman/listinfo/pytest-dev
