Date: Mon, 24 Nov 1997 13:49:54 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: DJ Delorie <dj AT delorie DOT com>
cc: djgpp-workers AT delorie DOT com
Subject: Possible enhancements on v2.02
Message-ID: <Pine.SUN.3.91.971124134758.23713f-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


I'd like to hear opinions on the following before I even try to make
some of them happen for v2.02:

     1) Is it a good idea to turn LFN support on Windows 95 *before*
        DJGPP.ENV is read.  (It will be turned off if DJGPP.ENV says
        so.)  This should take care of those who set
	DJGPP=c:/FubaricallyLongName/djgpp.env.

     2) Should the startup code automatically switch to unixy sbrk
        when we run on NT?  If there are cases where this would be
        contrary to user's intentions, we could have a flag that
        disables this.  (I assume that in most cases, unixy sbrk is
        the way to go on NT.)

     3) Currently, dosexec.c assumes a program name can have only one
        dot, and therefore doesn't try to look for `foo.bar.exe' when
        the app passes "foo.bar".  I think this is wrong under LFN.  I
        think under LFN we need to unconditionally look for the
        argument with all possible executable extensions before we
        look for it without an extension.  The changes in dosexec are
        minor, but I wonder whether they could break anything.

     4) How about testing the magic signature for ELF and other known 
	values, before calling DOS Exec function (and crashing) in 
	dosexec.c?