search

mspass-team / mspass

Massive Parallel Analysis System for Seismologists
https://mspass.org
BSD 3-Clause "New" or "Revised" License
28 stars 11 forks source link

Fix typos incorrectly using plural of file #499

Closed pavlis closed 5 months ago

pavlis commented 5 months ago

This branch fixes a typo using the plural in the storage_mode argument. Code in database.py and docstrings of distributed.py incorrectly used the plural "files" instead of "file".

pavlis commented 5 months ago

@wangyinz this is failing for a mysterious reason on an import error.

wangyinz commented 5 months ago

That error is due to the updated moto module, which was out today, that changed the api. You can see the change here: https://github.com/getmoto/moto/blob/5.0.0/CHANGELOG.md. I fixed that accordingly. However, now the distributed test also failed...

wangyinz commented 5 months ago

So, the new error is at line: https://github.com/mspass-team/mspass/blob/f72037521d4a70c6047a8b59c95f6488ae7cfe33/python/mspasspy/db/database.py#L7394

which complains: 

File "/opt/hostedtoolcache/Python/3.10.13/x64/lib/python3.10/site-packages/mspasspy/db/database.py", line 7394, in _construct_ensemble
    ens_tmp, abortions = self._load_ensemble_file(this_mdl)
TypeError: Database._load_ensemble_file() missing 3 required positional arguments: 'object_type', 'dir', and 'dfile'

and that makes sense because the _load_ensemble_file is defined as: https://github.com/mspass-team/mspass/blob/f72037521d4a70c6047a8b59c95f6488ae7cfe33/python/mspasspy/db/database.py#L7187

I don't understand how this was not failed before. Could you please take a look? @pavlis

pavlis commented 5 months ago

Ah. I see where this is coming from now. This is happening in a new section of save_data for ensembles that adds a "feature" that was not present before. The docstrings in the chain explain it somewhere but the idea is to efficiently save data in an ensemble with mixed storage modes. The main idea i to sort out all data going to a single file so we can use the faster ensemble write written in C. This particular problem is an example of something I noted in recent meetings. This write of Database has untested features. This is one of them - ensembles with mixed storage modes.

I may fork a completely new branch to address this and create the tests I should have made in the first place. The good news is the bug won't effect most workflows and there is a (slow) workaround to not set storage_mode to anything but the default