On Thu, Sep 12, 2019 at 02:16:34AM +0200, Tomas Vondra wrote:
> I still think using procnum 0 and passing the data through fn_expr are not
> the right solution. Firstly, traditionally the amprocs are either required
> or optional, with required procs having low procnums and optional starting
> at 11 or so. The 0 breaks this, because it's optional but it contradicts
> the procnum rule. Also, what happens if we need to add another optional
> amproc defined for all AMs? Surely we won't use -1.
>
> IMHO we should keep AM-specific procnum and pass it somehow to the AM
> machinery.
The latest review has not been addressed, and this was 7 weeks ago.
So I am marking the patch as returned with feedback.
--
Michael