Re: date & touch: timegm/mktime checks

On Sun, Jun 21, 2026 at 09:17:16AM -0600, Theo de Raadt wrote:
> These look correct.
> 
> But they look like useless chicken scratches, which an uneducated developer
> might delete later, don't they?
> 
> We also don't just want to to this in our tree, but for the idiom to be
> copied far and wide.
> 
> I think the docs and the landed diffs will fail to educate the outside
> community on using this pattern, and the manual page example should be
> improved and then idiomatically copied into all the places of use.
> 
> Index: time/ctime.3
> ===================================================================
> RCS file: /cvs/src/lib/libc/time/ctime.3,v
> diff -u -p -u -r1.55 ctime.3
> --- time/ctime.3        21 Jun 2026 08:16:35 -0000      1.55
> +++ time/ctime.3        21 Jun 2026 15:10:33 -0000
> @@ -406,7 +406,7 @@ did not change from \-1, an error has oc
>  .Bd -literal -offset indent
>  struct tm tm;
>  \&...
> -tm.tm_wday = -1;
> +tm.tm_wday = -1;       /* sentinel for error */
>  if (mktime(&tm) == -1 && tm.tm_wday == -1)
>         goto error;
>  .Ed
> 
> Which means all the tm.tm_wday = -1; in the tree should get that
> comment.

Sure, I'm fine with that.