Improved CF decoding

[mankoff]

• [X] Closes #2304 - but only for my specific use case.
• [x] Tests added

The comments above this line state, "so we just use a float64" but then it returns np.float32. I assume the comments are correct. Changing this also fixes a bug I ran into.

Note that currently, _choose_float_dtype returns float32 if the data is float16 or float32, even if the scale_factor dtype is float64.

[mankoff]

Note - I also have not run the "Running the performance test suite" code in https://xarray.pydata.org/en/stable/contributing.html - I assume changing from float32 to float64 would impact performance. I can run that if suggested.

[mankoff]

I'm reading more in https://github.com/pydata/xarray/blob/2a5686c6fe855502523e495e43bd381d14191c7b/xarray/coding/variables.py and I'm confused about some logic:

https://github.com/pydata/xarray/blob/2a5686c6fe855502523e495e43bd381d14191c7b/xarray/coding/variables.py#L271-L272

pop_to does a pop operation - it removes the key/value pair. So line 1 above will remove add_offset from attrs if it exists. The second line then checks for "add_offset" in attrs which should always be False.

I think this is happening based on inspecting with the debugger.

Furthermore, the fix I implemented in this Pull Request which returns np.float64 fixes my bug, but only because this bug exists. My dataset has add_offset, so the lines I changed:

        if not has_offset:
            return np.float64

should not run, but do run because of this issue.

[dcherian]

Sorry for dropping this @mankoff How can we move forward here?

[mankoff]

Hi @dcherian - I dropped this because I went down a rabbit hole that seemed very very deep.

Xarray has written 10s (100s?) of tests that touch this decoding function that make assumptions that I believe are incorrect after a careful reading of the CF spec. I believe the path forward will take some conversation before coding, so perhaps this should be moved to an issue rather than a pull request? A big decision is if the decode option strictly follows CF guidelines. If so, then a lot of tests need to be changed (for example, to follow the simple rule of [scale_factor and add_offset] must both be of type float or both be of type double).

Enforcing this would probably break xarray backward compatibility for writing files. I assume that that may be OK and there are processes to handle this (start with 'deprecation' warnings, then eventually throw errors?). There are also likely many NetCDF files that are not standard compliant and we need to decide how to read them.

Furthermore, the CF conventions are themselves not very clear, and possibly ambiguous. I started a conversation here: https://github.com/cf-convention/cf-conventions/issues/374 on this, but that is also unresolved at the moment. The CF convention mentions int and float, but not how many bytes those are. What happens when a files is written & packed on one architecture and read & unpacked on another?

[mankoff]

A bit more detail about the existing tests that don't match the CF spec. Per the spec, scale_factor and add_offset should be of the same type. That causes tests throughout https://github.com/pydata/xarray/blob/main/xarray/tests/test_coding.py and https://github.com/pydata/xarray/blob/main/xarray/tests/test_backends.py to fail, because:

https://github.com/pydata/xarray/blob/13c52b27b777709fc3316cf4334157f50904c02b/xarray/tests/test_coding.py#L112-L113

There is 1 test in test_coding, and 9 tests in test_backends that use mixed types. That's a tractable number I can fix.

In addition, the expected dtype returned by many of the tests does not match (my interpretation of) the expected dtype per the CF spec.

I am concerned that this is a significant change and I'm not sure what the process is for making this change. I would like to have some idea, even if not a guarantee, that it would be welcomed and accepted before doing all the work. I note that a recent other large PR to try to fix cf decoding has also stalled, and I'm not sure why (see #2751)

[dcherian]

A big decision is if the decode option strictly follows CF guidelines.

I think our general position is to be flexible on what we can read because there are many slightly non-compliant files out there.

Xarray has written 10s (100s?) of tests that touch this decoding function that make assumptions that I believe are incorrect after a careful reading of the CF spec.

Some of these might just be for convenience and some might be checking that we are flexible in what we can read.

This following test should be preserved so we can read those files (#4631):

 @pytest.mark.parametrize("scale_factor", (10, [10])) 
 @pytest.mark.parametrize("add_offset", (0.1, [0.1])) 

Enforcing this would probably break xarray backward compatibility for writing files.

Do we not enforce that scale_factor and add_offset are of the same dtype on write? If so, we should consider that a bug and fix it.

I am concerned that this is a significant change and I'm not sure what the process is for making this change.

I think the way to move forward would be to figure out the smallest change that would fix (or even improve) #2304 and move on. We have a 30-minute bi-weekly meeting (#4001) that you're welcomed to attend and raise specific questions. The next one is Oct 26 at 9.30am Mountain Time

[dcherian]

We should figure out how to express some of this understanding as tests (some xfailed). That way it's easy to check when something gets fixed, and prevent regressions.