bug#46974: 28.0.50; [nativecomp] wrong-type-argument number-or-marker-p with native compiled code only

[Andrea Corallo]

Pip Cet <pipcet@gmail.com> writes:

> On Sun, Mar 7, 2021 at 7:41 PM Andrea Corallo <akrl@sdf.org> wrote:
>> Pip Cet <pipcet@gmail.com> writes:
>> > On Sun, Mar 7, 2021 at 3:24 AM Erik Hetzner <egh@e6h.org> wrote:
>> >> This seems to have fixed the issue. I built from c60f2f4 and I don’t
>> >> see the issue. Thank you so much!
>> >
>> > I'm being bold and closing this, as I'm pretty sure it was due to the
>> > bug we fixed. Please reopen if necessary.
>>
>> The fact that you feel sure about this make me suspecting that you may
>> already have a reproducer for this :)
>
> Nah, I'm probably just over-confident :-) . I did reproduce this, but
> only in a live session.

Mine was a good try :D

> Perfectly open question: how would you feel about adding LAP
> reproducers for cases like this one, where it's hard to get at the
> Lisp source? I think I've got the reproducer LAP code here, and it
> shouldn't be too hard to write comp--native-compile-lap, but if you'd
> prefer not to go that way I'm perfectly okay with it, too.

I'm open to the idea of having a new way to stress the code in the
testsuite, I'm a little doubtful this worth of, well certainly for this
specific case where the patch is clearly correct.

What's your feeling about this?

  Andrea