APR "should" have a way to set the timestamp. I'm not sure how portable that
is. If it isn't portable, then there will be a #define that can be used to
test whether it is supported on that platform. For those platforms, we could
fall back to a different mechanism.
And yes, I also know that I need to move apr_file_copy() over at some point,
too... running out of time before leaving Thursday...
Let's say that we record the time stamp in the admin file _for_now_. That
will be our fallback code. When we get the APR functions to set the time,
then we can optimize the mechanism.
Cheers,
-g
On Tue, Sep 12, 2000 at 03:09:35PM -0500, Karl Fogel wrote:
> Manoj Kasichainula <manoj@collab.net> writes:
> > > If not, another possibility is to record the timestamp textually as
> > > CVS does. (Sigh).
> >
> > Well, I would assume subversion records the time that each delta was
> > committed. So won't you have the timestamp recorded anyway?
>
> By the principle of using available information efficiently, I'd
> prefer to just set the local working file and its text-base
> doppelganger to have the same timestamp, and then compare those.
>
> But if that's not possible to do reliably, then recording the working
> file's up-to-date timestamp textually in an administrative file (and
> simply ignoring the text-base file's timestamp) is the next best
> thing.
>
> (This is regarding updates & local mods, not commits.)
--
Greg Stein, http://www.lyra.org/
Received on Sat Oct 21 14:36:08 2006