Message-Id:	<199904251935.TAA109626@out1.ibm.net>
From:	"Mark E." <snowball3 AT usa DOT net>
To:	djgpp-workers AT delorie DOT com
Date:	Sun, 25 Apr 1999 15:36:28 -0400
MIME-Version:	1.0
Subject:	Re: fsext patches for dup and dup2
In-reply-to:	<199904251638.MAA09874@envy.delorie.com>
References:	<Pine DOT SUN DOT 3 DOT 91 DOT 990425190529 DOT 7199G-100000 AT is> (message from Eli Zaretskii on Sun, 25 Apr 1999 19:09:59 +0300 (IDT))
X-mailer:	Pegasus Mail for Win32 (v3.01d)
Reply-To:	djgpp-workers AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

> The only problem I can think of off-hand is that dup/dup2 are usually used
> just before running a child program, to set up stdin/stdout, and that just
> won't work for fsext handles.
> 

True, it won't work for fsext allocated handles. But when fsext is used to 
keep state information about a actual handle (e.g. for emulating the 
pipes), then adding fsext code to dup & dup2 is essential for the 
emulation to work properly (e.g. to know when the file being used to 
emulate the pipe can be safely deleted). For cases where a fsext 
allocated handle is being used (for Unix style access to serial ports, 
memory mapping as the fsext docs mention), you wouldn't want to 
dup2 it to stdin/stdout anyway. Perhaps I should write up for the docs 
the above uses and nonuses of fsexts for dup and dup2 if the patch is 
eventually accepted?

Mark

--- 
Mark Elbrecht, snowball3 AT usa DOT net
http://snowball.frogspace.net/

- Raw text -