These are chat archives for frictionlessdata/chat
delete()it and then
write()the new data into it. This is problematic in the case the write fails and we're left without a valid dataset. Even if the write succeeds, we still have a down time of the data for the writing period (could be many minutes).
delete_allparameter to the
write()method, which deletes all rows and writes them back in the same transaction. It's not very efficient but might work.
update_byparameter, which allows you to define a set of columns as a 'key' - new rows are tested to see if an existing row with the same key already exists. If it does, data is updated. If it doesn't a new row is added.
datasetlibrary uses for it
keysparameter - https://dataset.readthedocs.io/en/latest/api.html#dataset.Table.update