Martijn van Oosterhout wrote:

On Sun, Jan 22, 2006 at 09:04:14AM -0500, Andrew Dunstan wrote:
If you don't like relying on file order to resolve this, appropriate
use of %prec would have the same effect (just like for operator
precedence). The output file tell you which way bison went.

If we allow shift/reduce or reduce/reduce conflicts, debugging future development becomes more difficult. Right now we have the nice property that if you see one of those you know you've done something wrong (and using the expect directive isn't really a good answer, and only applies to shift/reduce conflicts anyway).

But that's the point of the %prec directive. To force bison to choose
one or the other, thus removing the warning... For an ambiguity that
only appears in one statement, it seems a better solution than upgrade
SET to a new class of identifier.

Quite so. We already use %prec extensively. All I was pointing out was that using file order isn't an acceptable option.

Presumably the effect in this case would be to prevent anyone from using SET as an alias unless there was a preceding AS.



---------------------------(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

Reply via email to