Simon Riggs <simon(at)2ndquadrant(dot)com> writes:
> My only fear is your reminder at a later date that we can't add feature
> X because of constant indexes and 8.4 is going to be all about indexes.
Well, if there's actually a reason to forbid it at some point, we can
reconsider. Right now there is no such reason. Furthermore, I rather
imagine that if there were a reason, the restriction it would require
would be a bit different than the one under discussion here. If there's
something we want to do that cannot work with index expressions that
happen to be constants, it seems unlikely that adding a partial index
predicate would suddenly make it start to work.
regards, tom lane