On Fri 24 Apr, 12:24 +0200, Christian Kastner <c...@debian.org> wrote:
I have to disagree on that. I'd agree if PyPI were hosting the official version, but this appears to be provided by a third party. This unofficial version already conflicts with upstream, so future conflicts and issues cannot be ruled out. What it the third party deviates even more, or the maintainer loses interest and lets this version go stale? It's not inconceivable to rely on a third party for bindings (I myself do that for libfann, for example), but in caes of doubt, I believe we need to follow upstream. In any case, the official version has certainly seen far more widespread use than the PyPI version (the former being around for a decade, the latter for a few months), so we need to be aware that changing this might break a lot of existing code out there...
I understand. In an ideal world, upstream would just take cover PyPI packaging and the whole mess would be over... But you are also right that the official version has been around far longer than the very young PyPI port...
[sorry for not putting BTS in CC. I will bounce my emails to BTS right now so that this discussion doesn't get lost.
Thanks! Tiziano -- debian-science-maintainers mailing list debian-science-maintainers@alioth-lists.debian.net https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/debian-science-maintainers
