Re: bug or simply not enough stack space?

Frank van Vugt  writes:
>> I think I might have been overenthusiastic in trying to free resources
>> during a subtransaction abort.  Try this patch and see if you notice
>> any bad side-effects.

> All examples I had that crashed and burned, now work correctly and/or bail out 
> correctly where needed.

> No side-effects noticed.

Well, it had some :-(.  I've committed a better version of that patch,
which will be in 8.4.1.  However, if you're not seeing any serious
memory leakage then you could keep using the patch you have for the
time being.

			regards, tom lane