Tom Lane wrote:
While all this reasoning is perfectly OK on its own terms, it ignores
the precedents of SQL identifier handling. Maybe we should revisit the
question of whether the labels are identifiers.
If we do that can we still cache the values in the syscache? My
impression from what you and TomD said was that it would be at least
more difficult. If it's possible I'm all in favor.
(Side note: ISTM there is a pretty good case to move the truncation code
from the lexer to the parser where it can be invoked according to
context. Maybe something to think about next cycle.).
cheers
andrew
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
