Andrew Dunstan <andrew@dunslane.net> writes:
> On 11/08/2014 12:37 AM, Michael Paquier wrote:
>> Well, yes :) I missed that. Note that I am leaning to Robert's
>> direction as well to do a clear separation... Now if the final
>> consensus is different, then let's use the patch attached that puts
>> the SQL functions to builtins.h, and the rest in quote.h.
> I am unlcear about what the consensus is on this, and don't have strong
> feelings either way. Do we need a vote? It's not of earth-shattering
> importance, but my slight inclination would be to do the minimally
> invasive thing where there is disagreement.
Well, the minimally invasive thing would be to reject the patch
altogether. Do we really need this?
In a quick look, the patch seems to result in strictly increasing the
number of #include's needed, which ISTM is not a positive sign for a
refactoring, especially given the number of files it hits. If there
had been some #include's removed as well, I'd be happier.
regards, tom lane