Peter Eisentraut wrote:
> Bruce Momjian writes:
>
> > Peter Eisentraut wrote:
> > > Bruce Momjian writes:
> > >
> > > > I assume the GEQO results he is seeing is only for a tests, and that the
> > > > macro version of newNode will help in all cases.
> > >
> > > Well are we just assuming here or are we fixing actual problems based on
> > > real analyses?
> >
> > What is your point? He is testing GEQO, but I assume other places would
> > see a speedup from inlining newNode.
>
> That's exactly my point: You're assuming.
We know GEQO is faster, and the 8% is enough to justify the inlining.
Whether it speeds up other operations is assumed but we don't really
care; the GEQO numbers are enough. The code bloat is minimal.
> Of course everything would be faster if we inline it, but IMHO that's the
> compiler's job.
Yes, but we have to help it in certain cases, heap_getattr() being the
most dramatic. Are you suggesting we remove all the macros in our
header file and replace them with function calls?
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073