Re: Possible copy and past error? (\usr\backend\commands\analyze.c)

Em sex., 27 de mar. de 2020 às 20:49, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> escreveu:

> Ranier Vilela <ranier(dot)vf(at)gmail(dot)com> writes:
> > Can someone check if there is a copy and paste error, at file:
> > \usr\backend\commands\analyze.c, at lines 2225 and 2226?
> > int num_mcv = stats->attr->attstattarget;
> > int num_bins = stats->attr->attstattarget;
>
> No, that's intentional I believe. Those are independent variables
> that just happen to start out with the same value.
>
Neither you nor I can say with 100% certainty that the original author's
intention.

>
> > If they really are the same values, it could be changed to:
>
> > int num_mcv = stats->attr->attstattarget;
> > int num_bins = num_mcv;
>
> That would make it look like they are interdependent, which they are not.
>
> That's exactly why, instead of proposing a patch, I asked a question.

> > To silence this alert.
>
> If you have a tool that complains about that coding, I think the
> tool needs a solid whack upside the head. There's nothing wrong
> with the code, and it clearly expresses the intent, which the other
> way doesn't. (Or in other words: it's the compiler's job to
> optimize away the duplicate fetch. Not the programmer's.)
>
I completely disagree. My tools have proven their worth, including finding
serious errors in the code, which fortunately have been fixed by other
committers.
When issuing this alert, the tool does not value judgment regarding
performance or optimization, but it does an excellent job of finding
similar patterns in adjacent lines, and the only thing it asked for was to
be asked if this was really the case. original author's intention.

regards,
Ranier Vilela