Re: master b467bb5 4/4: Minimize ‘equal’ calls in (delete x vector)

[Pip Cet <pipcet@gmail.com>]

On Sat, Aug 15, 2020 at 7:42 PM Paul Eggert <eggert@cs.ucla.edu> wrote:
> On 8/15/20 12:22 PM, Pip Cet wrote:
>
> > I'd prefer the attached: this is probably not worth worrying
> > too much about, so go with something straightforward?
>
> Thanks for the quick review and patch. I came up with something a bit simpler
> and installed the attached.

Looks correct, but establishes a rather questionable precedent of
putting Lisp_Objects in SAFE_ALLOCA'd space. That's okay in this
specific case, for a number of reasons, but it's bound to be copied by
someone in circumstances where it's not :-)

> It does bug me a bit that GCC generates bad code for 'NILP (Fequal (...))' here.
> I've been meaning to add a bool function equal (...) that would simplify the C
> source a bit and presumably let GCC do better, but that can wait.

What's GCC doing, for you? I see that it doesn't eliminate the
potentially dead store to kept[n] here, but other than that the code
looks okay to me.