842 dump and load scalars of sessions #843
Conversation
larray/inout/hdf.py
Outdated
| res = LGroup(key=key, name=name, axis=axis) | ||
| except KeyError: | ||
| scalar_key = key | ||
| attrs = store.root._v_attrs |
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
There was a problem hiding this comment.
From the description of the "obj" argument of the tables.Array class (at http://www.pytables.org/usersguide/libref/homogenous_storage.html#tables.Array), it seems like Python scalars might be natively supported by pytables...
| axis = read_hdf(filepath_or_buffer, attrs['axis_key']) | ||
| res = LGroup(key=key, name=name, axis=axis) | ||
| elif _type in {cls.__name__ for cls in _supported_scalars_types}: | ||
| res = pd_obj.values |
There was a problem hiding this comment.
I wonder if you should not finish #761 before this PR, because supporting scalars via Pandas is not actually needed. We will need to support them in the "direct"/#761 path anyway and have to support Arrays, Axis and Groups on both (for backward compatibility) be we can skip scalars via Pandas (that would mean a few lines of code fewer to maintain).
There was a problem hiding this comment.
Now it is there, let's finish this. I don't know when I will go back to work on PR #761 (partly because I don't like to work with pytables and prefer h5py but my intuition tells me we need pytables for our case).
There was a problem hiding this comment.
I thought h5py had the same features than pytables with a different API, except for querying (and optionally indexing) tables. And since we wouldn't need either feature, I would think using h5py would not be a problem. That said I never used h5py so I don't know for sure.
In either case, I wouldn't use h5py anyway because pandas relies on pytables and since we will have to keep the current code (using pytables indirectly) for backward compat for a while, depending on both pytables and h5py would be weird.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
|
OK with the last commit? |
|
sure |
|
So, OK to merge? |
…ession object (hdf5 + pickle formats)
…ore updating conda
alixdamman
force-pushed
the
842_dump_and_load_scalars_of_sessions
branch
from
1f4f6db to
d8a1561
Compare
|
Thanks for the review. |
No description provided.