=?utf-8?Q?Micha=C5=82_K=C5=82eczek?= <michal@kleczek.org> writes:
> When implementing a GiST consistent function I found the need to cache pre-processed query across invocations.
> I am not sure if it is safe to do (or I need to perform some steps to make sure cached info is not leaked between
rescans).
AFAIK it works. I don't see any of the in-core ones doing so,
but at least range_gist_consistent and multirange_gist_consistent
are missing a bet by repeating their cache search every time.
> The comment in gistrescan says:
> /*
> * If this isn't the first time through, preserve the fn_extra
> * pointers, so that if the consistentFns are using them to cache
> * data, that data is not leaked across a rescan.
> */
> which seems to me self-contradictory as fn_extra is preserved between rescans (so leaks are indeed possible).
I think you're reading it wrong. If we cleared fn_extra during
rescan, access to the old extra value would be lost so a new one
would have to be created, leaking the old value for the rest of
the query.
regards, tom lane
