Load+save things other than filepaths, and specifically support netCDF4.Dataset-s #5024
Conversation
|
NOTE: some of this will need re-visiting, |
| @@ -587,7 +587,7 @@ def load_cubes(file_sources, callback=None, constraints=None): | |||
| warnings.warn("{}".format(e)) | |||
|
|
|||
| # Perform any user registered callback function. | |||
| cube = run_callback(callback, cube, cf_var, cf.filename) | |||
| cube = run_callback(callback, cube, cf_var, file_source) | |||
There was a problem hiding this comment.
I think this makes more sense.
I believe it is consistent with what we do for UM "structured loading"
see "Use of load callbacks" in structured_um_loading api docs
-- though in this case it is the filename arg that is different to usual, rather than the field one.
|
Just re-raising this, and put into Iris 3.5 project, since some such feature will be important to the proposed solutions for #4994 |
|
Closing in favour of #5214 Which now targets main, and required a lot of rebasing ! |
Replaces the iris-load changes part of #4835,
So, re-presented as a separate issue which should be much simpler.
TBD: reform + add the other parts of that logic, in subsequent PRs.
NOTE:
I think that, whilst we are targetting a feature-branch, it isn't generally appropriate to add whatsnew-s.
So, we can sort out how to announce+explain it all when we merge back the FB.
Deferring the changes also probably avoids a whole lot of trivial conflicts whenever we do merge main-to-FB or vice versa.