Conversation
bc2c63f to
c061487
Compare
83219a3 to
fd8afb6
Compare
|
So this should be all good to go, but there are now three test failures: Which is interesting since I've not touched any test files. Maybe something to do with the environment creation! |
trexfeathers
left a comment
There was a problem hiding this comment.
This is coming together nicely! Couple of final observations.
5752267 to
4d405e1
Compare
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #5184 +/- ##
=======================================
Coverage 89.37% 89.37%
=======================================
Files 89 89
Lines 22426 22426
Branches 5379 5379
=======================================
Hits 20043 20043
Misses 1637 1637
Partials 746 746 ☔ View full report in Codecov by Sentry. |
|
This is getting closer. I just have to make change to the benchmarks and test them locally. |
69e92f8 to
6b34302
Compare
|
Benchmarks ran, so this should now be ready! |
|
Since everything is in a single commit I think I'll need to go back through all the changes - I can't remember which bits I did and didn't review back in March. So this will need to wait until after the sprint. |
|
Thanks! no rush at all! |
trexfeathers
left a comment
There was a problem hiding this comment.
OK, I've been through this with fresh eyes, including lots of trial tox runs locally, and I think it makes sense. The Tox world seems a lot slicker than the Nox world, even if I did need to learn a new syntax 👍
So my only outstanding caution is concerning dependency on tox-conda, since it's such a small project. See below.
| cache_period: ${{ env.CACHE_PERIOD }} | ||
| env_name: ${{ env.ENV_NAME }} | ||
| install_packages: "cartopy nox pip" | ||
| install_packages: "cartopy tox'<4'" |
There was a problem hiding this comment.
I feel a little uncomfortable that this remains necessary over 6 months later (see tox-dev/tox-conda#163).
tox-conda looks in need of more maintainers, are we comfortable being dependent on it? @lbdreyer @bjlittle
There was a problem hiding this comment.
There was a problem hiding this comment.
Having thought more, that was ill-informed, since Tox still offers neat inheritance tricks. But if we stick with tox-conda, we need to be comfortable sticking with the pin as well. Of course being pinned kinda undermines the idea of adopting something more popular.
Thanks @ESadek-MO and @HGWright for conversation on this topic.
There was a problem hiding this comment.
Also leaves us to potential security vulnerabilities:
|
Looks like this didn't get recorded, but discussions between @SciTools/peloton in December concluded that we should investigate how easy it would be to maintain our own tox-conda plugin, with just the features that we need. @pp-mo expressed an interest in this, I don't know how it has progressed. We accepted that this is less good than maintaining the real, public plugin. But we also agreed that it could be the pragmatic approach since a private plugin would take less maintenance and make us less vulnerable to outside forces. @HGWright highlighted that if we got a proof-of-concept together then we could potentially start a conversation on the tox-conda repo to see how many others are maintaining their own plugin and discuss pooling resources. This matters for this PR, as I am increasingly reluctant to approve tox-conda becoming a dependency. |
|
|
WIPThis is now ready for reviewThis switches our use of nox to tox, re #4806