Date: Fri, 04 Apr 1997 17:33:17 +1200
From: physmsa AT cantua DOT canterbury DOT ac DOT nz (Mr M S Aitchison)
Subject: Re: YESCHAR
To: OpenDOS AT delorie DOT com
Reply-to: M DOT Aitchison AT phys DOT canterbury DOT ac DOT nz
Message-id: <199704040533.RAA21557@cantua.canterbury.ac.nz>


I do think it is important to be able to change the "Yes" character
(and a handful of others).  As somebody else pointed out there are some
situations where the initial letter can mean the exact opposite of what
you want.  And, given that OpenDOS is trying to be international, and
supports seperate tasks, it is potentially risky to have one task
change this while other tasks are running, so the best approach is to
keep separate memory areas per task.  That isn't a huge memory
requirement (I can imagine wishlists where DOS would bloat, of
course, so I still appreciate your comments).

In the case of the switch character, and the option of allowing "/dev",
only a byte per variable needs to be assigned, not the length of the
whole keyword name (assuming the command.com does something sensible with its
own code).

The solution to useability questons is often not bloat (although that
often comes up as teh first option), but simply good design.  For
example, a minicomputer system I worked on years ago had a seperate
messages file which could be substituted, and *most* system commands
used these messages.  In the DOS situation I'd imagine that there would
be a series of bytes accessed by the DOS call with AH=37h, AL would
just need a shift right to get the offset. Supplementing that could be
a series of DOS calls to get a standard message by number (which would
read a text file given in some system area unknown to the programmer)
or accept a yes/no response.  This isn't new to operating systems,
doesn't cause bloat, and does lead to good programming practices.  You
can always fall back on a defined constant array in the program if the
DOS isn't advanced enough.

Mark.