As discussed on Slack:
The alternative way is, we start storing query uids in task.Result.
[4:54 PM]
Not all queries would have pagination, in fact very few branches will.
[4:55 PM]
So, we can just rewrite task.Result for those branches.
[4:55 PM]
That’d still be an overall cheaper operation than using the slower PBs to FBs.
[4:55 PM]
Then we don’t need to look at task.Query to interpret task.Result.
[4:56 PM]
Most task.Results would stay as they were retrieved from network. Only some need to be changed, and for those, we just rewrite them.
[4:56 PM]
Pagination of say 100 results or so, essentially makes that a very cheap operation anyway.
[4:56 PM]
when we could have millions of results in one task.Result.