Re: Cache invalidation bug in RelationGetIndexAttrBitmap()

Tomas Vondra <tv@fuzzy.cz> writes:
> On 14.5.2014 22:29, Andres Freund wrote:
>> I'm afraid it's more in the year range from what i've seen. I.e. not
>> practical.

> Yeah, that wouldn't be very practical. I'll try to run it though and if
> it'd run more than a few days, I'll switch it to CLOBBER_CACHE_ALWAYS.

I think it might be interesting to try running CLOBBER_CACHE_RECURSIVELY
on some subset of the regression tests.  But I'm not sure which subset,
nor whether the buildfarm scripts could readily be coaxed to do that.
Let's see how far you get with a full run before you get bored ...
        regards, tom lane