On 22/08/2018 18:49, David G. Johnston wrote: > What others have done doesn't change the situation that has arisen for > PostgreSQL due to its implementation history.
What others have done seems relevant, because the whole reason these questionable interfaces exist is to achieve compatibility across SQL implementations. Otherwise you can just make a native SQL call directly.
It seems to me that if we don't make it possible to call a function or a procedure using
the same mechanism the drivers will have to make a choice which one to implement.
That said the path of least resistance and regression for the drivers would be to not implement
calling procedures through each respective drivers mechanism. I would think given the importance of
this work it would be a shame not to make it easy to use.
I also agree with David that driver writers made the best out of the situation with functions and we are now asking for the server to dual purpose the call command.
Is there a technical reason why this is not possible ?