Flow data results in a merged dataset is blank

[ukanga]

The flow data results endpoint returns blank for a merged dataset form- https://api.ona.io/api/v1/flow-results/packages/{UUID}/responses whereas the data endpoint - https://api.ona.io/api/v1/data/{pk}.json - would return data.

There should be data returned from the flow data results endpoint.

Aha! Link: https://ona.aha.io/features/PROD-233

[DavisRayM]

Merged Datasets / MergedXForms currently do not have a UUID. We might need to add the field in order to not go against the FLOIP Spec.

I feel like the migration necessary for this may break backwards compatibility. Should I go forward with this approach ?

CC: @ivermac @pld

[pld]

Why would it break backwards compatibility?

[DavisRayM]

Why would it break backwards compatibility?

Pre-existing Merged Datasets won't have the field and I feel like the UUID field should not allow null values. I think that would cause a lot of issues.

[DavisRayM]

The other approach would be to allow null values on the field. But I'm not so sure how to go about updating and adding uuid's for Merged Datasets created before this change is implemented.

[pld]

Why would it break backwards compatibility?

Pre-existing Merged Datasets won't have the field and I feel like the UUID field should not allow null values. I think that would cause a lot of issues.

Can we use a data migration to fill this field? An require that it not be null

[DavisRayM]

Why would it break backwards compatibility?

Pre-existing Merged Datasets won't have the field and I feel like the UUID field should not allow null values. I think that would cause a lot of issues.

Can we use a data migration to fill this field? An require that it not be null

Confirmed with @ivermac seems like setting a default would fill the field. And from this post here, seems like its possible to set a uuid as default.

I'll go ahead with this approach.

[DavisRayM]

Researched a bit into the FLOIP spec as suggested here. Noted down the following:

The FLOIP spec supports the following Question types: multiple_choice_one/select_one, multiple_choice_many/select_many, numeric, open, text, image, video, audio, geo_point, date, time, datetime

The multiple_choice_many and multiple_choice_one seem to have been replaced with select_one and select_many... The spec lists multiple_choice_* here, but here it's select_one and select_many

The 'name' metadata should be on the same level as 'title' according to this and this

The other change already identified in the conversation on #1773 is that the 'id' returned should be UUID4 (hex+digits) this would possibly require us to change how we set uuid's on onadata.. Optionally we can use uuid.UUID function to convert the hex values back to the full uuid

I think this changes should be in a separate issue all together, if they're needed.... thoughts ?

CC: @ukanga @ivermac @WinnyTroy @pld

[ukanga]

The other change already identified in the conversation on #1773 is that the 'id' returned should be UUID4 (hex+digits) this would possibly require us to change how we set uuid's on onadata..

I think we do not necessarily need to change that but we could. From how we store it you can get the dashed version:

>>> import uuid
>>> uuid.UUID('a94dc0c6f4bb4cea95af8df3969c67bc')
UUID('a94dc0c6-f4bb-4cea-95af-8df3969c67bc')
>>>

[ukanga]

Please create issues for the others so that we can make the changes to conform to specification.

[DavisRayM]

Please create issues for the others so that we can make the changes to conform to specification.

Created an issue for the other changes