You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
The return out of BaseBuilder.parse_access_ncfile is currently a long tuple of assorted file information. This makes it difficult to remember which item is which during debugging & testing.
Describe the feature you'd like
We should consider if parse_access_ncfile should return a dict instead. This will have the advantage of making the return more human-legible. (It will also simplify the parser functions of the various builders, which effectively all rebuild the tuple return into a dict anyway.)
Describe alternatives you've considered
Nil.
Additional context
I'm about to add another issue (#187) to suggest an expansion of the testing for builders, which will be made easier by this feature.
The text was updated successfully, but these errors were encountered:
The changes in #212 should cover this - I'd initially thought that we'd want to have separate flags in the catalogue for data and coordinate variables, so I factored it out as a way of keeping track of changes. It turned out to be unnecessary to do that, but I agree the long tuple is difficult to keep track of & creates a lot of unnecessary context.
I used a dataclass, not a dict, however - let me move a lot of the repetitive stuff inside the class. Given pydantic is used pretty heavily upstream in intake-esm, I wonder whether we might want to use a pydantic dataclass? It might be handy to enforce the extra validation?
Is your feature request related to a problem? Please describe.
The return out of
BaseBuilder.parse_access_ncfile
is currently a long tuple of assorted file information. This makes it difficult to remember which item is which during debugging & testing.Describe the feature you'd like
We should consider if
parse_access_ncfile
should return adict
instead. This will have the advantage of making the return more human-legible. (It will also simplify theparser
functions of the various builders, which effectively all rebuild the tuple return into adict
anyway.)Describe alternatives you've considered
Nil.
Additional context
I'm about to add another issue (#187) to suggest an expansion of the testing for builders, which will be made easier by this feature.
The text was updated successfully, but these errors were encountered: