On Thu Apr 8, 2021 at 3:56 PM BST, Xabier Oneca -- xOneca wrote:
> Hi,
>
> > On Thu Apr 8, 2021 at 9:24 AM BST, Rasmus Villemoes wrote:
> > > Why not go all the way and start using utimensat() and do nanoseconds?
> > > That's the resolution presented in struct stat, and what most file
> > > systems store - without that, it's not possible to accurately replicate
> > > one file's timestamps on another. Also, POSIX nowadays specifies touch
> > > in terms of utimensat()
> > > <https://pubs.opengroup.org/onlinepubs/9699919799/utilities/touch.html>.
> > >
> > > Rasmus
> > Actually, it seems utimensat() is a function specified in SUSV4
> > that would make FEATURE_TOUCH_SUS3 somewhat misleading
> > Perhaps it should be renamed to FEATURE_TOUCH_SUSV4 if that is going to
> > be the approach taken
> >
> > That would also make the code further down more messy, as you'd need
> > more #if to cover a version using utimes(), and a version using
> > utimenstat()
> >
> > What do you think?
>
> Actually, I wold remove (l)utimes and use only utimenstat, without
> #ifdefs.
>
> And about the SUSv3 options... I was thinking about renaming that
> "group" of options to something like "fancy" (eg.
> FEATURE_TOUCH_FANCY), like in other applets.
>
> Thus things will be clearer. No?
>
> Cheers,
>
> Xabier Oneca_,,_
That approach sounds good - I think if that was the case, then there
wouldn't really be have to be a need for a seperate option to use
nanoseconds, it can just be part of the fancy options.
_______________________________________________
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
