search

lincc-frameworks / tape

[Deprecated] Package for working with LSST time series data
https://tape.readthedocs.io
MIT License
12 stars 3 forks source link

Test EnsembleFrame Serialization Methods #312

Open wilsonbb opened 9 months ago

wilsonbb commented 9 months ago

Dask supports various serialization methods for its DataFrames (see here), and for the EnsembleFrame hierarchy we should validate that we still have appropriate round trip serialization, both in terms of the underlying data as well as our metadata (_Frame.label, _Frame.ensemble, _Frame.dirty)

Note that one approach we already do is that we override _args to avoid an issue present in dask-geopandas. Our testing can help us determine if this is needed or if the linking to the issue in our comments can be expanded or removed.

wilsonbb commented 9 months ago

Quick notes on priority:

Having end-to-end testing of serialization of the new dataframes seems worth doing but is relatively low priority and low-to-moderate effort.

We might get this as part of any effort to save and reload an ensemble which we have discussed doing, but it seems worth keeping this issue open to preserve context on the issue seen in dask-geopandas.