Date: Tue, 30 Jun 1998 10:14:16 +0300 (IDT)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: "Salvador Eduardo Tropea (SET)" <salvador AT inti DOT gov DOT ar>
cc: djgpp-workers AT delorie DOT com
Subject: Re: Windows 9X/NT file times
In-Reply-To: <m0yqj1e-000S3uC@inti.gov.ar>
Message-ID: <Pine.SUN.3.91.980630101339.6274E-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Mon, 29 Jun 1998, Salvador Eduardo Tropea (SET) wrote:

First, thanks a lot for running the tests on your machine.

> I got:
> 
> foo1   06/29/1998 15:39:13.280  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> foo2   06/29/1998 15:39:21.030  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> foo3   06/29/1998 15:39:25.140  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> foo4   06/29/1998 15:39:26.510  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> foo5   06/29/1998 15:39:27.620  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> foo6   06/29/1998 15:39:28.610  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> foo7   06/29/1998 15:39:29.520  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> foo8   06/29/1998 15:39:30.560  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> foo9   06/29/1998 15:39:31.960  06/29/1998 00:00:00.000  06/29/1998 15:47:50.000
> 
> Surprise! Creation in the future!

Sorry, I don't see where's the creation in the future in this output.
The creation time is the first time printed, so in your example it is
*before* the last-write (aka modification) time, which is the last of
the three times printed by the program.  (Programs which look at the
time file was modified, like GNU Make, use the modification time.)

> 4) Now: touch foo[1-9]:
> 
> foo1   06/29/1998 15:51:12.380  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> foo2   06/29/1998 15:51:17.940  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> foo3   06/29/1998 15:51:19.870  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> foo4   06/29/1998 15:51:20.720  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> foo5   06/29/1998 15:51:21.540  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> foo6   06/29/1998 15:51:22.410  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> foo7   06/29/1998 15:51:23.280  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> foo8   06/29/1998 15:51:24.200  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> foo9   06/29/1998 15:51:25.130  06/29/1998 00:00:00.000  06/29/1998 15:51:44.000
> 
> Again! the files are in the future. Looks like touch is messing the
> creation time.

I don't see how is this test significant.  `touch' doesn't change the
creation time at all, it only changes the modification time and the
last-access time, even on Windows 9X.  And since the last-access time
is only a date, you will only see the change in the modification time,
unless you wait past midnight.

So the above results seem okay to me: `touch' has changed the
modification time, but left the creation time intact.

To compare the creation and modification time, you need to delete the
files before running `touch', so that both times are (theoretically)
the same.  At least on my system, this always causes the modification
time to lag behind the creation time, which is expected behavior since
the modification time has 2-second granularity.