Re: review: More frame options in window functions

I wrote:
> I started looking at this patch. I believe that we should commit the
> AggGetMemoryContext API function --- *not* the window context
> management changes that you included here, but only the API abstraction
> --- to be sure that that gets into 9.0 so that third-party aggregate
> functions can start relying on it instead of looking directly at the
> AggState or WindowAggState node. The rest of the patch might or might
> not make it, but we can at least help people future-proof their code.

I have committed that little part. I revised the function API to be

/* AggCheckCallContext can return one of the following codes, or 0: */
#define AGG_CONTEXT_AGGREGATE 1 /* regular aggregate */
#define AGG_CONTEXT_WINDOW 2 /* window function */

extern int AggCheckCallContext(FunctionCallInfo fcinfo,
MemoryContext *aggcontext);

so that it would be conveniently usable in places that just want to
check aggregate-ness and don't need to fetch a memory context; and
with the thought that maybe someday there would be more than two
possible call contexts.

regards, tom lane