In #4930 (comment), we discovered that PostgREST returns a 206 response on an unfiltered endpoint when using count=estimated or count=planned.
Of course, an unfiltered endpoint always returns the full response, so can never be "partial". We can't avoid the problem entirely, i.e. with a filter that returns all rows, the same thing can still happen, so I'm unsure how useful it'd be to special case that.
However, we should check that at least we can't return a wrong Content-Range header, when the estimated number of rows is lower than the actually returned number. I assume this can happen, too, but I haven't seen it, yet.
In #4930 (comment), we discovered that PostgREST returns a 206 response on an unfiltered endpoint when using count=estimated or count=planned.
Of course, an unfiltered endpoint always returns the full response, so can never be "partial". We can't avoid the problem entirely, i.e. with a filter that returns all rows, the same thing can still happen, so I'm unsure how useful it'd be to special case that.
However, we should check that at least we can't return a wrong Content-Range header, when the estimated number of rows is lower than the actually returned number. I assume this can happen, too, but I haven't seen it, yet.