This is the mail archive of the
cygwin
mailing list for the Cygwin project.
Re: more ctime bugs
On Feb 13 04:30, Eric Blake wrote:
> Corinna Vinschen <corinna-cygwin <at> cygwin.com> writes:
> > I'll update Cygwin to set ctime in close and link. Link
> > is special since it doesn't involve using any explicit file descriptors,
> > so it's a bit unclear where to set the flags inside Cygwin to get that
> > right. Using close() seems a good way to have ctime set for write()
> > as well as open(O_TRUNC).
>
> I see the new has_changed flag in the 20050211 snapshot. But you still have to
> add a call to touch_ctime() within the stat() family of calls if has_changed is
> set, in order to comply with the required semantics; stat and lstat are not
> allowed to return out-of-date timestamps.
You know that this contradicts the target to maintain speed on write?
It's not done with adding a call to touch_ctime() to fstat, because
that only affects the application which also has written to the file.
If I read SUSv3 correctly, any call to stat/fstat/lstat on the file
by any application would require to update the timestamp. That's no
fun. I'm willing to simply ignore this.
Corinna
--
Corinna Vinschen Please, send mails regarding Cygwin to
Cygwin Project Co-Leader mailto:cygwin@cygwin.com
Red Hat, Inc.
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/