[reporter="edan", created="Wed, 22 Mar 2006 12:39:16 -0500 (GMT-05:00)"]
operations involving recordtype failing in validator e.g. select (1,2)=(1,1) from values(1)
It seems like it fails in different places and reasons in the farrago validator compared to the saffron one.
------- Comment #5 From Julian Hyde 2005-11-08 14:47 [reply] -------
Operations on record types are better handled using code generation (at sql-to-rex translation time). For example,
[reporter="edan", created="Wed, 22 Mar 2006 12:39:16 -0500 (GMT-05:00)"] operations involving recordtype failing in validator e.g. select (1,2)=(1,1) from values(1)
It seems like it fails in different places and reasons in the farrago validator compared to the saffron one.
------- Comment #5 From Julian Hyde 2005-11-08 14:47 [reply] -------
Operations on record types are better handled using code generation (at sql-to-rex translation time). For example,
(x, y) = r
should translate to
x = r.field1
and y = r.field2
with a liberal sprinkling of null semantics.