On 11/12/2010 02:25 PM, Clark C. Evans wrote:
> On Fri, 12 Nov 2010 21:10 +0000, "Dann Corbit" wrote:
>> If (for access) the single table seems simpler, then
>> a view can be used.
>
> Even if you "partition" the columns in the instrument
> over N tables, you still can't query it in a single
> result set. The limit is quite deep in PostgreSQL
> and extends to tuples, including views and in-memory
> query results.
>
> I find that partitioning does work, but it requires extra
> care on the part of the application developer that really
> shouldn't be necessary.
>
> Best,
>
> Clark
>
A single questionnaire has 1600 questions. Fine. That all of those
questions/answers are definitive to a single analyzable entity boggles
the mind. There are no sections like "if previous answer was no, skip
the next section" or "choose any of the following values which apply"?
No demographics block for example? I see the point of "getting it all
back simply" but there's also the cost of always having all the other
stuff when I'm only interested in (hypothetical) sections 1, 5, and 7.