On Wed, 9 Feb 2022 at 15:46, Ethan Furman <et...@stoneleaf.us> wrote:
> On 2/9/22 6:59 AM, Barney Gale wrote: > > > I've now submitted a PR that adds an initial underscore-prefixed > implementation of this class: > > Why the iniitial underscore? What could change between now and the future? > A few things: Firstly, I'd like to give PyPI package maintainers the opportunity to play with the underscore-prefixed API and submit feedback, prior to us freezing the interface. They're already doing naughty things subclassing the (internal, recently removed) pathlib._Accessor class. I think this feedback might identify the need for new OSError subclass representing EINVAL for readlink() implementations, or a new user-friendly os.stat_result constructor for use in stat(). Or perhaps not :-) Secondly, there's still a couple of rough edges in the PurePath implementation, particularly around certain operations being delegated to a mix-in class (the "flavour"). I'd like to bring those into PurePath before we unleash this on the world, but it can be worked around for the time being. Thirdly, I've yet to address the problem of how we supply things like sockets or zipfile.ZipFile objects, and persist them to child objects generated by iterdir() etc. It can be done already by creating a further subclass with the state assigned as a class attribute, but that's a bit weird I think!
_______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-le...@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/QOIDZYXZXQE27NG3J7UFHB6HQQKYIF4V/ Code of Conduct: http://python.org/psf/codeofconduct/
