From: sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id: <10212131422.AA16306@clio.rice.edu>
Subject: Re: proposed putpath.c patch
To: djgpp-workers AT delorie DOT com
Date: Fri, 13 Dec 2002 08:22:56 -0600 (CST)
In-Reply-To: <3791-Fri13Dec2002105929+0200-eliz@is.elta.co.il> from "Eli Zaretskii" at Dec 13, 2002 10:59:30 AM
X-Mailer: ELM [version 2.5 PL2]
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

> > From: sandmann AT clio DOT rice DOT edu (Charles Sandmann)
> > Date: Fri, 13 Dec 2002 01:54:23 -0600 (CST)
> > 
> > This patch makes the /dev directory work for me in most cases.
> 
> I have some concerns about this (but I'm not sure they are grave
> enough to object to the change):

It's a first pass for comments.  I'd like more testing.

>  - this makes _putpath rather slow for Unix-style device names

True, only using unix-style device names which aren't unix devices.
And we should only call _putpath on the open, so one additional
interrupt in that case would probably be unnoticeable.

It also makes _put_path slow for all files in a /dev directory,
but that could be fixed with a simple caching.

The workaround there is to document this, and suggest don't use
things like /dev/com1 or /dev/clock$ (don't add the /dev/ at all)
if you expect it to work with a /dev directory present.

>  - if /dev exists and we are on W2K/XP, won't this break DOS devices
>    such as /dev/LPT1 ?  I seem to recall those systems have problems
>    with creating the illusion any device exists in any directory.

We fixed all that with the LFN patches (I think.  I hope...)
I tried some various quick operations with /dev/com1 and they seemed
to behave the same.  It needs more testing.

The next question is: how many programs use /dev/dos_device_name 
syntax?  Then, how many of them use that syntax on a system with a
/dev directory present?

>  - doesn't function 43h have problems on W2K/XP?

There was some different behavior for devices compared to Win9x family,
but nothing important here (seeing if a directory exists).