Re: Curing plpgsql's memory leaks for statement-lifespan values

On 22 July 2016 at 07:02, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> In
>
> https://www.postgresql.org/message-id/tencent_5C738ECA65BAD6861AA43E8F@qq.com
> it was pointed out that you could get an intra-function-call memory leak
> from something like
>
> LOOP
> BEGIN
> EXECUTE 'bogus command';
> EXCEPTION WHEN OTHERS THEN
> END;
> END LOOP;

...

>
> Another answer is to invent a third per-function memory context intended
> to hold statement-lifespan variables.
>
>
I've wanted this in the past and been surprised we don't have it, so +1
from me.

I don't think a few MemoryContextAlloc's are too ugly.

I suggest that in cassert builds, or maybe just CACHE_CLOBBER_ALWAYS, the
context is reset after each call even when not cleaning up after an error.
That'll help catch mistakes and leaks.

--
Craig Ringer http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services