bug#39944: 27.0.90; JIT Stealth timer errors

[Eli Zaretskii]

> Cc: 39944@debbugs.gnu.org
> From: Paul Eggert <eggert@cs.ucla.edu>
> Date: Sat, 7 Mar 2020 10:02:28 -0800
> 
> > (the comments in that part of the code can really use some
> > enhancement; they currently seem to target only experts in both time
> > handling and GMP: many macros whose names don't explain what they do,
> > and plethora of calls to libgmp functions that have no comments
> > whatsoever).
> 
> Although comments in that area no doubt could use improvement, I'd rather not 
> see comments like this:
> 
>        /* Add TM_YEAR_BASE to mpz[0].  */
>        mpz_add_ui (mpz[0], mpz[0], TM_YEAR_BASE);

There are quite a few of similar comments there, and they actually
help, because they save us from consulting the GMP manual on every
step.

But I really meant that stuff like this lacks a comment:

      hz = make_integer_mpz ();
      mpz_swap (mpz[0], *iticks);
      ticks = make_integer_mpz ();

And also functions like timespec_ticks, lisp_time_hz_ticks, and
lisp_time_seconds, which don't have a single comment describing how
they do their thing.  Following that code with the purpose of tracking
some specific result is not for the faint at heart.  I think adding a
comment here and there might make that easier.

> > +  (when (numberp time)
> > +    (setq time (timer-relative-time nil time))
> > +    (or (consp time)
> > +        (setq time (time-convert time 'list))))
> 
> This would catch some problems but not all, as the real bug here is in the 
> code 
> (or (consp time) (error "Invalid time format")) which occurs a few lines 
> later. 
> As near as I can tell this later code is both wrong and unnecessary. It's 
> wrong 
> because it's no longer true that only conses are time values. It's 
> unnecessary 
> because the immediately following (timer-set-function timer function args) 
> call 
> checks the validity of TIME. On the off chance that a validity check is still 
> helpful (because we don't want to create garbage?) I installed the attached 
> patch.

Thanks.

> But it might be better in master to remove the "Invalid time format"
> check entirely.

I'm fine with removing that test, if we are sure that invoking
run-at-time with something utterly un-timely, like a symbol or a a
string that cannot be a valid time description, will trigger an error
(presumably from timer-set-time).