TCTracks read_ibtracs_netcdf() AttributeError about climada_python

[tovogt]

Ah, this might be caused by the fact that the package xarray introduced the string interface only starting from version 0.13. Okay, this means you have to upgrade xarray to 0.13 at least: conda install xarray=0.13.

@emanuel-schmid Can we update the xarray version from 0.12.1 to 0.13 in the conda requirements?

[emanuel-schmid]

Ah, this might be caused by the fact that the package xarray introduced the string interface only starting from version 0.13. Okay, this means you have to upgrade xarray to 0.13 at least: conda install xarray=0.13.

@emanuel-schmid Can we update the xarray version from 0.12.1 to 0.13 in the conda requirements?

Yes, I will at least try.

[simonameiler]

I got it running now. A simple restart of all programms was enough. Sorry to have bothered you with these basic problems. I'll close the issue with this.

[mmyrte]

I presume you're on the development branch; it's changed quite a bit from master. Either you could try using master if you don't depend on develop, or you could check the results in tc_tracks.py from line 210 onward. Simply read the netcdf in a notebook and play around with it. sid is the storm ID, maybe it doesn't get set correctly.

[simonameiler]

Your presumption is right; I am using develop. Unfortunately; I depend on it. But I'll just sit back and relax because I don't trust myself fixing this bug.

[tovogt]

I'm afraid, I have no idea what might have caused your problem. I can't reproduce it with any of the last few versions of the develop branch. Please test again with the latest checkout from develop. If the problem persists, please also post the value of the basin variable and the commit hash of your HEAD.

[simonameiler]

The problem persists. Maybe my git rebase action before pulling develop caused the problem.

commit hash of my HEAD:
b282880

basin variable: 'NA'

tracks_IB = TCTracks()
tracks_IB.read_ibtracs_netcdf(basin='NA')

[mmyrte]

Maybe I can latch on regarding the dependencies: since conda tries to resolve it all in one go, I would have suggested trying an all-out upgrade. I think we can safely rely on the tests in that scenario. The last full upgrade seems to be from sometime in mid-2019, based on the pinned versions.

[emanuel-schmid]

I presume you're on the development branch; it's changed quite a bit from master. Either you could try using master if you don't depend on develop, or you could check the results in tc_tracks.py from line 210 onward. Simply read the netcdf in a notebook and play around with it. sid is the storm ID, maybe it doesn't get set correctly.

Please do not use master, just because develop fails. In the end develop will become master and if you cannot solve the problem on develop the problem is not solved.

[mmyrte]

Please do not use master, just because develop fails. In the end develop will become master and if you cannot solve the problem on develop the problem is not solved.

Sure, I didn't mean "fix the bug in master" - I merely meant that the bug was introduced by develop. If Simona could have written her code without depending on develop, she could simply have continued working...

About the dependency upgrade: does it warrant a separate issue to keep track of it? Is it desirable in the first place?

[emanuel-schmid]

Definitely desirable but not quickly done. The dependency carousel in CLIMADA is quite a beast and because of the cross-dependencies timing is crucial. There is no point in making this an issue for it will never be solved.
The general task to update them on a regular base is part of my maintenance service.
When a particular version causes problems, this needs its own issue.

[simonameiler]

Ah, this might be caused by the fact that the package xarray introduced the string interface only starting from version 0.13. Okay, this means you have to upgrade xarray to 0.13 at least: conda install xarray=0.13.

@emanuel-schmid Can we update the xarray version from 0.12.1 to 0.13 in the conda requirements?

I updated xarray according to Thomas' suggestion and also updated netcdf4. Now, I still seem to have a problem at another end:

tracks_dict = dict()
tracks_IBTrACS = init_tc_tracks_IBTrACS('NA')
Traceback (most recent call last):

  File "<ipython-input-17-ea06de5baa8e>", line 2, in <module>
    tracks_IBTrACS = init_tc_tracks_IBTrACS('NA')

  File "/Users/simonameiler/Documents/WCR/Scripts/TC_IB_K_G_his_1.py", line 167, in init_tc_tracks_IBTrACS
    tracks_IB.read_ibtracs_netcdf(year_range=(1980, 2017), basin=basin, correct_pres=False)

  File "/Users/simonameiler/Documents/WCR/CLIMADA_develop/climada_python/climada/hazard/tc_tracks.py", line 226, in read_ibtracs_netcdf
    ds = xr.open_dataset(fn_nc)

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/site-packages/xarray/backends/api.py", line 499, in open_dataset
    filename_or_obj, group=group, lock=lock, **backend_kwargs

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/site-packages/xarray/backends/netCDF4_.py", line 389, in open
    pop_to(attributes, encoding, "least_significant_digit")

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/site-packages/xarray/backends/netCDF4_.py", line 335, in __init__
    if format is None:

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/site-packages/xarray/backends/netCDF4_.py", line 398, in ds
    dsvars = FrozenDict(

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/site-packages/xarray/backends/netCDF4_.py", line 392, in _acquire
    encoding["original_shape"] = var.shape

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/contextlib.py", line 112, in __enter__
    return next(self.gen)

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/site-packages/xarray/backends/file_manager.py", line 183, in acquire_context
    @contextlib.contextmanager

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/site-packages/xarray/backends/file_manager.py", line 195, in _acquire_with_cache_info
    """Acquire a file, returning the file and whether it was cached."""

  File "/Users/simonameiler/opt/anaconda3/envs/climada_env/lib/python3.7/site-packages/xarray/backends/lru_cache.py", line 52, in __getitem__
    with self._lock:

TypeError: descriptor '_lock' for 'LRUCache' objects doesn't apply to 'LRUCache' object

[tovogt]

I'm running my code with netcdf4 version 1.4.2. That's the one in the CLIMADA requirements file. Is there any reason why you decided to upgrade netcdf4?

[simonameiler]

I upgraded it to 1.4.2 as well. The reason why I chose to update it was that I ran into a bottleneck error that in my understanding was associated to my previous version of netcdf4. But I'm happy to go hunt down the error on my side.