Tom Lane wrote:
> argue that there was a regression. It's certainly a performance bug
> though: nobody would expect that giving a query *more* work_mem would
> cause it to run many times slower.
I wouldn't be that surprised - otherwise it'd just be hard-coded to
something large.
Especially since earlier in the thread:
> The example is *not* particularly slow if you leave work_mem at default.
which makes me think it's arguably not quite a bug.