Currently One._check_filesystem supports the datasets input being a slice without the eid index. This is important because by default indexing a single level returns a view without that level (c.f. one._cache.datasets.xs(eid, level='eid', drop_level=False)):
If the eid level was assured, we could drop the session_path column (which is cumbersome to generate on Alyx and accounts for ~13% of the table size) and do joins instead (see https://github.com/int-brain-lab/ONE/issues/134). This might be slightly less performant, especially if the eid index is missing and has to be fetched again from the original table.
Currently
One._check_filesystem
supports the datasets input being a slice without the eid index. This is important because by default indexing a single level returns a view without that level (c.f.one._cache.datasets.xs(eid, level='eid', drop_level=False)
):If the eid level was assured, we could drop the session_path column (which is cumbersome to generate on Alyx and accounts for ~13% of the table size) and do joins instead (see https://github.com/int-brain-lab/ONE/issues/134). This might be slightly less performant, especially if the eid index is missing and has to be fetched again from the original table.