-
Notifications
You must be signed in to change notification settings - Fork 31
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Current logic in the frontend prevents using the default filename #153
Comments
agreed, I think something in the logic has been broken. @huard and @aulemahal any thoughts on this? PRs are always welcome! |
I agree that this would make sense! I don't think it ever worked, though, seems to me that the current behaviour was introduced when It think it would be a matter of setting |
I am testing the solution proposed. Not unexpectedly, it breaks this test:
Because that combination of arguments now compute and save the weights to the default file. I am not sure about how to proceeded in order to not to break this behavior. Maybe add a special case with filename="default"? |
The filename logic dates back to the time ESMF wrote the weights to file, and xESMF had to read them from disk. About two years ago, ESMF introduced a mechanism to share the weights in memory. The current code is the result of trying to support both the older file-based and memory-based approaches without breaking legacy code. I'm not sure I would want xESMF to automatically read from a default file name that is not directly visible from the docstring. It feels error-prone. |
Current logic in the frontend does not let me to use the default filename
I think it should try to load the weights from default filename instead of raising an error. Am I missing something?
The pattern used to build the default filename is good and useful. I can find some time to write a PR if you want.
The text was updated successfully, but these errors were encountered: