Bug#955015: python3-xarray: Backend api engine grib broken
Alastair McKinstry
alastair.mckinstry at sceal.ie
Fri Mar 27 13:13:32 GMT 2020
Hi,
I shall forward to upstream. It does appear that setting engine=cfgrib
should not be necessary if the file is a grib file (by extension) and
cfgrib is present.
I Agree with your point on writable_stores / cfdisk.
Best regards
Alastair
On 27/03/2020 12:44, Fabrice Meyer wrote:
> Just adding a detail, to avoid messing with xarray code, you just need
> to call to_netcdf with an engine define in your parameters. It is just
> the default engine selection which appears to be broken on this version
> of xarray (and also in stable).
>
> Regards
>
>
> Le 26/03/2020 à 21:07, Fabrice Meyer a écrit :
>> Package: python3-xarray
>> Version: 0.15.0-3
>> Severity: important
>>
>> Dear Maintainer,
>>
>> I was trying to use to_netcdf() located in /usr/lib/python3/dist-packages/xarray/backends/api.py on a grib file without specifying the engine and it appears that the default engine selection end up on grib engine.
>>
>> So _get_default_engine calls _get_default_engine_grib as we have a grib file but this function always raise a ValueError even if cfgrib import succeed but anyway, there is a real issue with this function as it is returning nothing while _get_default_engine is expecting an engine.
>> If we try to, somehow, makes _get_default_engine_grib return cfdisk as engine, it will still fail as cfgrib is absent of the WRITEABLE_STORES bellow but even if it would have been present, cfgrib does not implement a writeable store.
>>
>> WRITEABLE_STORES: Dict[str, Callable] = {
>> "netcdf4": backends.NetCDF4DataStore.open,
>> "scipy": backends.ScipyDataStore,
>> "h5netcdf": backends.H5NetCDFStore.open,
>> }
>>
>> I'm quite new dealing with xarray but I think grib engine seems to be on work in progress or something and doesn't look ready yet.
>>
>> In order to get the job done, I tryed to make _get_default_engine_grib returning "netcdf4" based on observation on an older implementation of xarray (0.10.2-1). This "hack" worked but it is really dirty.
>>
>> Could you please dig a bit more and fix this issue?
>>
>> Best regards, and thank you for all job done maintaining theses packages.
>>
>>
>> -- System Information:
>> Debian Release: bullseye/sid
>> APT prefers stable-updates
>> APT policy: (500, 'stable-updates'), (500, 'unstable'), (500, 'testing'), (500, 'stable')
>> Architecture: amd64 (x86_64)
>>
>> Kernel: Linux 4.19.0-8-amd64 (SMP w/1 CPU core)
>> Kernel taint flags: TAINT_CRAP
>> Locale: LANG=fr_FR.UTF-8, LC_CTYPE=fr_FR.UTF-8 (charmap=UTF-8), LANGUAGE=fr_FR.UTF-8 (charmap=UTF-8)
>> Shell: /bin/sh linked to /bin/dash
>> Init: systemd (via /run/systemd/system)
>> LSM: AppArmor: enabled
>>
>> Versions of packages python3-xarray depends on:
>> ii python3 3.8.2-2
>> ii python3-numpy 1:1.17.4-5
>> ii python3-pandas 0.25.3+dfsg-7
>>
>> Versions of packages python3-xarray recommends:
>> ii python3-bottleneck 1.2.1+ds1-2+b1
>> ii python3-dask 2.11.0+dfsg-1
>> ii python3-h5netcdf 0.7.1-1
>> ii python3-netcdf4 1.5.3-1+b2
>> ii python3-zarr 2.4.0+ds-1
>>
>> Versions of packages python3-xarray suggests:
>> pn python-xarray-doc <none>
>> pn python3-cartopy <none>
>> ii python3-matplotlib 3.2.1-1
>> pn python3-pydap <none>
>> pn python3-rasterio <none>
>> ii python3-scipy 1.3.3-3
>> pn python3-seaborn <none>
>> ii python3-toolz 0.9.0-1
>>
>> -- no debconf information
--
Alastair McKinstry, email: alastair at sceal.ie, matrix: @alastair:sceal.ie, phone: 087-6847928
Green Party Councillor, Galway County Council
More information about the debian-science-maintainers
mailing list