should header content type matter when starting job with a stored dataset

[frantjc]

I'd wager no; if they sent the storage, then there's only one type (json or zip), and we just use that

If they're using the output of another job, then all types are there (as of writing, json and zip) and whatever we use should result in the same output. We can prioritize one if its faster than another, but I don't think the user needs to worry about what type we use.

[phish3y]

@frantjc I agree on everything you said, lets use this ticket to make sure the API doc reflects that the header content type doesn't matter if you're using an existing storage.

Also, there's the case of the lookups where the result is JSON only and not GeoJSON. You can't chain jobs on the job output of these tasks. We should reflect that in the API doc and code. No input-of, etc for those task types.

[phish3y]

There's a partial lie in the second paragraph above. The vecotor lookup actually does produce valid GeoJSON, so you could chain jobs on that. The raster lookup does not, however.

[frantjc]

per discussion on Discord:

• new Task column to differentiate between lookup and transformation tasks types
• new Storage column to differentiate between storage that is transformable (e.g. transformation task outputs), final (e.g. lookup outputs), unknown (e.g. user inputs that have not yet been used) and unusable (e.g. user inputs which have been indicated as bad data from a task)
• new Jobs cannot be made from Jobs whose Task type is a lookup or Storages whos status is final or unusable

[phish3y]

Also update API documentation

[frantjc]

created https://github.com/logsquaredn/geocloud/issues/195 from this discussion, changed from question => documentation as the question has been answered...