The Forum for Discussion about The Third Manifesto and Related Matters

You need to log in to create posts and topics.

Self partitioning a relation

Just a thought: does anyone see any value to partitioning a relation over a projection of itself? For example [maybe not a good example]:

Get suppliers with fewer than three shipments.

SP {S#} WHERE COUNT ( ‼SP ) < 3

Quote from Greg Klisch on April 1, 2020, 5:02 pm

Just a thought: does anyone see any value to partitioning a relation over a projection of itself? For example [maybe not a good example]:

Get suppliers with fewer than three shipments.

SP {S#} WHERE COUNT ( ‼SP ) < 3

Well, if it provides the correct answer to your question then that is the value, isn't it ?

Quote from Greg Klisch on April 1, 2020, 5:02 pm

Just a thought: does anyone see any value to partitioning a relation over a projection of itself? For example [maybe not a good example]:

Get suppliers with fewer than three shipments.

SP {S#} WHERE COUNT ( ‼SP ) < 3

Looks like a good example to me.

If you find the query readable and it generates the results you expect under all cases, then it's of value.

I'm the forum administrator and lead developer of Rel. Email me at dave@armchair.mb.ca with the Subject 'TTM Forum'. Download Rel from https://reldb.org
Quote from Greg Klisch on April 1, 2020, 5:02 pm

Just a thought: does anyone see any value to partitioning a relation over a projection of itself? For example [maybe not a good example]:

Get suppliers with fewer than three shipments.

SP {S#} WHERE COUNT ( ‼SP ) < 3

There was a long and meandering thread about Date's IMAGE operator !!. My take is that some shorthands around 'partitioning a relation over a projection of itself' would be useful, but that !! is not the most useful of those. That partitioning is exactly what the GROUP operator does.

SP_PART := SP GROUP { P#_QTY := {ALL BUT {S#}}}   // might have got the syntax wrong

SP_PART EXTEND : { PQ_COUNT := COUNT(P#_QTY); Q_SUM := SUM(QTY FROM P#_QTY)} WHERE PQ_COUNT < 3

Most of the examples for DIVIDE and friends focus on 'Suppliers that supply all parts' so need mention of two or more relvars. But I find it easier to think of GROUP each relvar individually to give RVAs then JOIN on the RVAs. Exactly you're partitioning each relvar over a projection of itself. (That's another reason I don't waste mental energy on DIVIDE.)