Run unittests on Python 3.12 #27
Conversation
mweinelt
force-pushed
the
tests
branch
4 times, most recently
from
8d646d6 to
53b428b
Compare
.github/workflows/test_package.yml
Outdated
| matrix: | ||
| python-version: [3.6] #, 3.7, 3.8] | ||
| pytorch-version: ["1.8.0", "nightly"] | ||
| python-version: ["3.9", "3.10", "3.11", "3.12"] |
There was a problem hiding this comment.
One version of Python only to save some compute.
There was a problem hiding this comment.
Okay, trying to figure out what versions actually work. Then I'll reduce the scope to one of those.
There was a problem hiding this comment.
3.12 seems to be the latest version that works, so I chose that.
The full test matrix takes 1m20s fwiw, which is probably fine.
mweinelt
force-pushed
the
tests
branch
3 times, most recently
from
75a08cc to
eb675fa
Compare
Pip compatible interface but 10-100x faster, thus saving on compute and giving faster response times. |
mweinelt
force-pushed
the
tests
branch
5 times, most recently
from
67a310a to
17d4cb4
Compare
- Test pytorch stable and nightly - Use uv's pip interface for faster dependency resolution - Migrate away from deprecated `py.test` executable - Convert various options to long-opts, makes them easier to understand
This is required for using uv I'm afraid, but the changes should increase the QoL of the life, since we can rely on a no-code approach to define the package. This is based on https://setuptools.pypa.io/en/latest/userguide/pyproject_config.html
|
Okay. Looks like tests to 2.5.1 are running on 3.9-3.12. The nightly torch tests fail during dependency resolution, because for some reason they want to include Time for dinner. |
Running on just one version was requested, as to reduce the compute we consume through the CI.
|
I had to pin to a newer numba version, to get a llvmlite version compatible with 3.10 and newer. Now all versions 3.9-3.12 are working with torch 2.5.1 and nightly. The |
mweinelt
changed the title
py.testexecutable