Does anyone have experience in using "SET ALL=" with multiple (cluster) joins?

I have a MFD composed of a 'narrow' table with 41 columns of application data and 24 columns of binary values used as 'foreign keys' to join this 'parent' record to each of 24 'child'/'look-up' tables (XRef tables in FOCUS lingo).

I expect to use "SET ALL=..." to control the acceptance of each 'parent' row when a joined 'child' row is present or missing, but I wonder if I may need to (or am able to) express a different ALL= value for each of the 24 'children'. Of course, in my case, all 24 joins are not likely to be in use at the same time because FOCUS optimizes away a join when a column in the 'child' table is not mentioned in a DEFINE, PRINT/SUM, WHERE, IF, or BY context. If no child column is required for the report, then the child table is not part of the SQL query.

At this time my question is ... How is the SET ALL= control implemented when there are joins to multiple tables?
Is the match/nomatch outcome of each join combined with all the other match/nomatch outcomes by "AND"ing them together or by "OR"ing them together to arrive at a final include/exclude decision?
If I have several joins active and some of them have matches while others do not, how will the final output differ for each of the three possible settings of ALL=?

My envoronment is v7.1 querying a MS/SQL relational database.

You don't want to set ALL=ON with multiple joins if your tables are relational - you want to explicitly set your inner and outer joins. Otherwise they will all be outer joins which is very inefficient.
You may not get any different results (IF your tables are in a relational database) for ALL=ON vs ALL=PASS.

Use either ALL=PASS or ALL=ON since this equivalent for RDBMS but both are sometime useless if you test on the joined TO structure or you may have to prepend a WHERE TOKEYFIELD EQ MISSING OR test-condition.

Heres how to put left outers in your clusters

SEGNAME=&SEGNAME, SEGTYPE=U, PARENT=&PARENT, CRJOINTYPE=LEFT_OUTER,

This message has been edited. Last edited by: hammo1j,