Paige (since you cc:'d me < s >)
AFAIK:

>I 'assume' that when I open a parent table with n child records, that the parent table and each child table is opened in seperate work areas and is buffered seperately.
yes.

>Question: If I am using Row buffering for all tables and if the pointer is moved off the parent record, committing the record in that table, will the child records be 'saved/written/committed' automaticaly?
Only if the record pointer in the child tables moves. Meaning, if there is a SET RELATION in place, I would operate under the assumption that the child's record pointer can move and program for that. But, then again, I wouldn't be inclinded to use row buffering on a child table (which I think we agree on?).

>If I am using Table buffering for all tables will no records be Saved untill/unless a TABLEUPDATE() is executed?
Never say never? Seriously, that's how it works. And keep in mind that the tableupdate() needs to be done on all the buffered tables. IOW, tableupdate only operates on one table. (I'm talking tables here and not views)

>If, as you suggest, I use Row buffering for the parent table and Table buffering for the child table and if the pointer is moved off the parent record, committing the record in that table, will the child records be 'saved/written/committed' automaticaly?
See above.