Date: Thu, 8 Jan 1998 10:51:06 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Robert Hoehne <robert DOT hoehne AT gmx DOT net>
cc: Vik Heyndrickx <Vik DOT Heyndrickx AT rug DOT ac DOT be>,
        DJGPP Mailing list <djgpp AT delorie DOT com>
Subject: Re: Rhide BUG: file dates under W95
In-Reply-To: <34B3F457.2C4F3E1C@gmx.net>
Message-ID: <Pine.SUN.3.91.980108105048.8789L-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Wed, 7 Jan 1998, Robert Hoehne wrote:

> Vik Heyndrickx wrote :
> > 
> > RHIDE 1.4 sometimes produces invalid file modification date/time, as
> > reported by SCANDISK.
> 
> Please try to create a reproducable example or explain it at
> least more detailed (what files are affected, what is the
> error message from SCANDISK, what have done and so on).

I second this.  I don't use SCANDISK much, but when I do, it never
reports anything like that.  Could it be that you sometimes boot your
machine into plain DOS (no Windows), and that SCANDISK complains about
files created/modified during plain DOS sessions?

> I do not so often use LFN=y but sometimes I do, and the only
> problem which I had with this and SCANDISK was, that I had sometimes
> reports about invalid ling filenames which are not longer valid
> but never about modification times.

This should also be investigated, IMHO.  LFN calls should never
produce invalid long names.

> The only thing I can imagine is, that W95 knows about two filetimes,
> the modifcation and access time (at least both can be stored in the
> directory entry)

Actually, Windows 95 knows about 3 times: creation, last access, and
modification time.  When you use DOS calls which don't support long
names, the first two times are nuked (replaced by zeroes), if memory
serves.