Build examples against Interchange 0.3.25 #46
Conversation
|
/generate-cache |
|
/regenerate-cache |
|
This repository caches data generated from notebooks stored in other repositories in the To regenerate the cache for this PR, create a comment on this PR consisting only of the string: |
|
A workflow has been dispatched to regenerate the cookbook cache for this PR.
|
|
A workflow dispatched to regenerate the cookbook cache for this PR has just finished.
If the workflow was successful, changes will only be visible in the ReadTheDocs preview after it has been rebuilt. |
|
While working on this, I noticed two sore spots:
I changed these upstream: openforcefield/openff-interchange@0cf4387 |
|
/regenerate-cache |
|
A workflow has been dispatched to regenerate the cookbook cache for this PR.
|
|
A workflow dispatched to regenerate the cookbook cache for this PR has just finished.
If the workflow was successful, changes will only be visible in the ReadTheDocs preview after it has been rebuilt. |
|
Okay, they're added back: https://openff-docs--46.org.readthedocs.build/en/46/examples.html |
There was a problem hiding this comment.
Great @mattwthompson, thanks for this! I'd like to avoid pip installing things that are available on Conda if at all possible - let me know if there's something wrong with that.
If you accept my suggestions and regenerating the cache again still works, consider this approval to merge!
devtools/conda-envs/examples_env.yml
Outdated
| - pip: | ||
| - git+https://github.com/openforcefield/openff-interchange.git@v0.3.25b1 |
There was a problem hiding this comment.
Can this be removed now that 0.3.25 has been released?
| - pip: | |
| - git+https://github.com/openforcefield/openff-interchange.git@v0.3.25b1 |
If you want a version of Interchange that's compatible with 0.3.25 installed with conda, but you want to take the examples from this tag, the best way to do that is to get ~=0.3.25 in the regular way and then set the tag in the GITHUB_REPOS global variable (l33-39 in globals_.py, change "openforcefield/openff-interchange" to "openforcefield/openff-interchange@v0.3.25b1"). I'd like to avoid pip installing things in this environment if possible.
There was a problem hiding this comment.
I just forgot to update this is all
devtools/conda-envs/examples_env.yml
Outdated
| @@ -14,7 +14,7 @@ dependencies: | |||
| - packaging | |||
| # Examples | |||
| - openff-toolkit-examples>=0.15.2 | |||
| - openff-interchange>=0.3.18 | |||
| - openff-interchange ~=0.3.25 | |||
There was a problem hiding this comment.
| - openff-interchange ~=0.3.25 | |
| - openff-interchange >=0.3.25 |
I think >= is actually the desired semantics - we want this environment not to need updating if an incompatible version of Interchange is released. There isn't any assumption that code will remain compatible with Interchange, because any code that uses Interchange comes either from the selected version of Interchange or a compatible version of a different package.
I can see that ~= is more appropriate for a downloadable environment for end-users, but the environment file is checked daily and isn't really designed for users to hold on to long term before they solve it. Users should re-download the new example and get the updated environment rather than use an old download.
What do you think?
There was a problem hiding this comment.
IIUC, ~= will pull down any 0.3.x that's released in the future but never pull down the 0.4.0 that I'm hoping exists at some point in the future.
In this case the difference may not be so big if 0.4.x doesn't break any examples, which is my hope, and if I'm going to be manually updating this anyway, which is more or less my expectation.
There was a problem hiding this comment.
Yeah ok totally fair. Sorry, I think I was worried that I'd be updating pins every time a minor release came out, but if it's your job to fix your examples then it's totally fair that you get to choose how to manage your releases!
|
/regenerate-cache |
|
A workflow has been dispatched to regenerate the cookbook cache for this PR.
|
|
A workflow dispatched to regenerate the cookbook cache for this PR has just finished.
If the workflow was successful, changes will only be visible in the ReadTheDocs preview after it has been rebuilt. |
|
A workflow has been dispatched to regenerate the cookbook cache for this PR.
|
|
A workflow dispatched to regenerate the cookbook cache for this PR has just finished.
If the workflow was successful, changes will only be visible in the ReadTheDocs preview after it has been rebuilt. |
|
A workflow has been dispatched to regenerate the cookbook cache for this PR.
|
|
QCArchive is down, not sure for how long |
|
A workflow dispatched to regenerate the cookbook cache for this PR has just finished.
If the workflow was successful, changes will only be visible in the ReadTheDocs preview after it has been rebuilt. |
|
I wonder if we could set up HOSTS to point |
|
It's usually somewhat flaky in CI, so not relying on MolSSI's server would be an improvement, but I wouldn't bet on that being a small amount of work |
|
/regenerate-cache |
|
A workflow has been dispatched to regenerate the cookbook cache for this PR.
|
|
A workflow dispatched to regenerate the cookbook cache for this PR has just finished.
If the workflow was successful, changes will only be visible in the ReadTheDocs preview after it has been rebuilt. |
|
Rebuilding and let's see ... |
No description provided.