Re: updating target at the end of recipe, after updating prerequisites, might not work?

[David Boyce]

On Mon, Nov 14, 2011 at 8:39 AM, Eli Zaretskii <address@hidden> wrote:>
Did you miss this portion of Coreutils' NEWS?
>    cp, install, mv, and touch now preserve nanosecond resolution on
>    file timestamps, on platforms that have the 'utimensat' and>    'futimens' 
> system calls.
Yes, I did miss that and I'm happy to know about it. For the record,
though, implementing this feature with traditional system calls would
be portable to many more platforms than relying on newly invented
ones. But the future direction is clear and I have no plan to pursue
it.
On Mon, Nov 14, 2011 at 9:25 AM, Paul Smith <address@hidden> wrote:
> Also, writing 0 bytes to a file does not (at least in my
> experiment) change its modtime.

I believe it depends on O_TRUNC. The command ">foo" always updates foo
but ">>foo" does not.

-David Boyce