On 06/30/2016 07:07 PM, Stefan Eissing wrote: > >> Am 30.06.2016 um 17:55 schrieb Yann Ylavic <[email protected]>: >> >> On Thu, Jun 30, 2016 at 5:38 PM, Stefan Eissing >> <[email protected]> wrote: >>> We now set exactly the same callback right before in line 709. If we had >>> more than one callback, we would not have to specify NULL, but restore any >>> previous callback there was, right? >> >> Actually NULL preserves the current callback, so once we've set one >> (or the default) at global/connection init time, if we use NULL >> everywhere else we can preserve it (patch proposed in my previous >> message). > > What I meant was: we would need to reset the callback that was in place > before line 709. The "NULL=ignore the parameter" will not suffice in case we > have more than one callback.
Fair enough. I didn't look at line 709 but only at line 804. So we handle things differently in both cases. I would like to see a consistent approach. But nothing terribly urgent or important. As said the patch works anyway in our case. Regards Rüdiger
