Re: Rethinking representation of partial-aggregate steps

David Rowley <david.rowley@2ndquadrant.com> writes:
> As for the above proposal, I do agree that it'll be cleaner with bit
> flags, I just really don't see the need for the AGGTYPE_* alias
> macros. For me it's easier to read if each option is explicitly listed
> similar to how pull_var_clause() is done, e.g:

That does not sound to me like it does anything to address the issue of
documenting which combinations of flags are sensible/supported.  I prefer
something like the enum approach I suggested further downthread.
        regards, tom lane