On Thu, Jun 6, 2024 at 3:05 PM Nathan Dunfield <nat...@dunfield.info> wrote:
>
>
>
> On Thursday, June 6, 2024 at 5:56:15 AM UTC-5 Dima Pasechnik wrote:
>
> Yet, pytest, a pip package, is installed and used rather regularly in Sage,
> and nobody gets hurt. It is found to be safe to install this particular
> package this way ("pip install pytest" does not affect anything in the Sage
> venv, that's why).
>
>
> Yet because the version of pytest wasn't pinned, things broke when pytest 8.*
> came out. This would not have happened if pytest had been installed as a
> wheel package. (As you say, pip packages can also be pinned to a particular
> version, though for a pure-Python package the only difference between wheel
> and a pinned pip is whether dependencies also have to be included.)
I never said that we must never pin versions of such packages.
OTOH Sage gets broken on macOS, mostly due to XCode updates, very often,
yet one does not demand pinning of macOS versions and XCode versions.
Compared to these, the incident with pytest 8 was very mild.
>
> Dima, under your proposal that "standard packages can be pip packages", what
> criteria would be used to decide whether (and how narrowly) to pin a
> particular pip package? Also, what criteria would be used to determine
> whether (and which) dependencies would be explicitly made Sage packages?
One can look at package dependencies and see what can and what cannot
be untangled from each other, and see.
Anyhow, I just want a possibility for standard pip packages (and
pytest fits the ticket).
If you were unwilling to grant such a possibility I would not want
spend time coming up with details.
Dima
>
> Personally, I support allowing Sage to use upstream binary wheels from PyPI
> rather than building from things source, but feel it is a mistake not to pin
> everything and explicitly list all dependencies, at least at first.
>
> Best,
>
> Nathan
>
> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sage-devel/fa7c82aa-ce78-471d-b5e2-35ff3a682b27n%40googlegroups.com.
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/sage-devel/CAAWYfq00h4L%2BktjSy2rrwquTPWvuBVmmkuWsOhbwvbO1BOARAg%40mail.gmail.com.
