Added validation of attrs before saving to netCDF files

[robintw]

This allows us to give nice errors if users try to save a Dataset with
attr values that can't be written to a netCDF file.

Fixes #911.

I've added tests to test_backends.py as I can't see a better place to put them. I've also made the tests fairly extensive, but also used some helper functions to stop too much repetition. Please let me know if any of this doesn't fit within the xarray style.

[shoyer]

Use dataset.variables (a dict of xarray.Variable objects) instead of separately checking data vars and coords

[robintw]

Thanks, I'll deal with these modifications.

Also, I'm seeing errors from TravisCI saying ValueError: cannot read or write netCDF files without netCDF4-python or scipy installed - what do I need to do to make sure these tests work on Travis? (They pass fine on my local machine)

[shoyer]

Take a look at the other netcdf tests -- they all use a decorator like
@requires_netCDF4, either on the class or the method. That ensures they are
skipped if required libraries for the functionality are not installed.
On Sun, Aug 28, 2016 at 2:43 PM Robin Wilson [email protected]
wrote:

Thanks, I'll deal with these modifications.

Also, I'm seeing errors from TravisCI saying ValueError: cannot read or
write netCDF files without netCDF4-python or scipy installed - what do I
need to do to make sure these tests work on Travis? (They pass fine on my
local machine)

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
#991 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/ABKS1qrd1mmelzaoL4omWV_TTLwAMeUSks5qkgEWgaJpZM4Ju5fU
.

[robintw]

I think I've dealt with all of these.

Refactoring the tests has made them a bit cleaner and safer due to not using mutation - however I still have quite a few repeated lines of code. I've also combined them all into one test with lots of asserts - as otherwise I couldn't keep the helper functions within the test without repeating them loads of times.

Does anything need adding to the docs for this? I'm assuming not as it's just making the code do more checking, it's not adding a feature as such?

[robintw]

The Travis builds seem to have all passed except for py27-cdat+pynio which has failed with a conda error that seems to be unrelated to these changes (Error: HTTPError: 403 Client Error: Forbidden for url)

[shoyer]

dataset.variables is a dictionary. So you can write:

for variable in dataset.variables.values():
    for k, v in variable.attrs.items():
        check_attr(k, v)

[shoyer]

This is almost there now.

I agree that we don't need to update the docs, but please do make note of it in "What's New".

[shoyer]

thanks!