Would be nice if import verified that each record array had N+1 items where N is the length of the channel_names array in the JSON.
Motivation: we had a user who forgot to include the timestamp in the record arrays, and so the datastore used the first value as the timestamp.
We'll need to discuss what action to take for something like the following:
{"channel_names":["foo", "bar", "baz"], "data":[
[1445624599,1,2,3],
[1445624600,4,5,6],
[7,8,9],
[1445624602,1,2,3]
]}Do we reject the entire import, or just ignore the one line with the incorrect number of fields? I think I'd probably vote for the latter, especially if we can report the number of ignored records in the response.
Would be nice if import verified that each record array had N+1 items where N is the length of the
channel_namesarray in the JSON.Motivation: we had a user who forgot to include the timestamp in the record arrays, and so the datastore used the first value as the timestamp.
We'll need to discuss what action to take for something like the following:
{"channel_names":["foo", "bar", "baz"], "data":[ [1445624599,1,2,3], [1445624600,4,5,6], [7,8,9], [1445624602,1,2,3] ]}Do we reject the entire import, or just ignore the one line with the incorrect number of fields? I think I'd probably vote for the latter, especially if we can report the number of ignored records in the response.