`default_version` arg. to `NetCDFRead.read` not exposed: remove?

[sadielbartholomew]

`default_version` arg. to `NetCDFRead.read` not exposed: remove?

[sadielbartholomew]

Is the answer to that to add those keys to the DEBUG logging from NetCDFRead?

I have done that already and so I guess we can indeed test by searching for the 'file_version' result in the log output string, and that would mean no need to expose the _scan_only keyword. Sounds like a plan! When I move on from ES-DOC work I'll do that for #296 and finish and open it for review. Thanks.

[sadielbartholomew]

There is another parameter I would like to expose in cf.read that is available in cf.NetCDFRead.read, namely _scan_only. This one is of course only intended for internal use, but I can't find an elegant-enough (non-contrived) way to query on the read variables registered (read_vars dict) without it, and I want to see those for testing e.g. in the associated PR #296.

That said, I think it could be a useful user-facing method - so I propose to upgrade it to one, possibly as part of the API review for cf v4.0.0.

[davidhassell]

Hi Sadie - what is the use case for querying the read_vars dict? (not doubting there is one - just wondered what you had in mind)

[sadielbartholomew]

In #296 I want to check that the correct read_vars["file_version"] and read_vars["UGRID_version"] (etc. for CFA version) get set, though perhaps we can just test that the ultimate Conventions property gets set, if we set it to the default e.g. CF-1.11 when an invalid value such as 1.6/1.7 is set?

[davidhassell]

Is the answer to that to add those keys to the DEBUG logging from NetCDFRead?