> > INSERT INTO t (a, b, c, d) VALUES ('aa', 'bb', (SELECT c, d FROM t > > WHERE <cond>)); > > with <cond> guaranteed to select exactly one row. > >I don't know of any DBMS where this would be valid.
It seems so. >INSERT INTO t(a, b, c, d) >SELECT 'aa', 'bb', c, d FROM t WHERE <cond>; My mistake: I had more than a single select but deleted them in my example to keep it short (but it then became trivial). What suprises me is that a select of 1 field yields a field (not a 1-tuple containing 1 field) but then a select of n > 1 fields yields a n-tuple, not a sequence of n fields. In Mathematica parlance: SameQ[{a,b,c,d}, {a,b,{c},{d}}] --> True, no Flatten required SameQ[{a,b,c,d}, {a,b,{c,d}}] --> False, a Flatten would be required Put in set theory, an element is equal to its singleton, but a set of n>1 elements remains a set and not a collection. That's fairly weird. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users