> select * from sometable where somefield IS NULL won't work because IS
is
> not a nomally indexible operator.
Ah, I didn't know that. So there is no real reason not to exclude null
values from all your indexes :). Reading Tom's recent comments
everything is clear now.
Instead of using your two index approach I prefer to:
create function nullidx(anyelement) returns boolean as $$ select $1 is
null; $$ language
sql immutable;
create index on t(nullidx(f)); -- etc
Merlin
