Re: mvcc catalo gsnapshots and TopTransactionContext

Andres Freund <andres@2ndquadrant.com> writes:
> On 2014-02-05 14:07:29 -0500, Tom Lane wrote:
>> I stuck such an Assert in ScanPgRelation, and verified that it doesn't
>> break any existing regression tests --- although of course the above
>> test case still fails, and now even without declaring an index.
>> 
>> Barring objections I'll go commit that.  It's a bit pointless to be
>> Asserting that catcache.c does nothing unsafe when relcache.c does
>> the same things without any such test.
>> 
>> (Alternatively, maybe we should centralize the asserting in
>> systable_beginscan or some such place?)

> I don't have a problem with sticking an additional assert elsewhere, but
> I think ScanPgRelation, systable_beginscan are a bit late, because they
> frequently won't be hit during testing because the lookups will be
> cached...

Oh, good point.  By analogy to the placement of the existing Assert in
SearchCatCache, the one for relcache should be in RelationIdGetRelation.

[ experiments... ]  OK, that passes regression tests too.  Good...
        regards, tom lane