Greetings,
* Tom Lane (tgl@sss.pgh.pa.us) wrote:
> Patch 0002 addresses several more-or-less-independent issues that are
> exposed by running the regression tests with patch 0001 activated:
>
> * Query.withCheckOptions fails to propagate through write+read, because
> it's intentionally ignored by outfuncs/readfuncs on the grounds that
> it'd always be NIL anyway in stored rules. This seems like premature
> optimization of the worst sort, since it's not even saving very much:
> if the assumption holds, what's suppressed from a stored Query is only
> " :withCheckOptions <>", hardly a big savings. And of course if the
> assumption ever fails to hold, it's just broken, and broken in a non
> obvious way too (you'd only notice if you expected a check option
> failure and didn't get one). So this is pretty dumb and I think we
> ought to fix it by treating that field normally in outfuncs/readfuncs.
> That'd require a catversion bump, but we only need to do it in HEAD.
> The only plausible alternative is to change _equalQuery to ignore
> withCheckOptions, which does not sound like a good idea at all.
I'm fine with this change (as I believe I've said before...). I agree
that it's just a minor optimization and shouldn't be an issue to remove
it.
Thanks!
Stephen
