Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84 SMI; site sun.uucp
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!genrad!decvax!decwrl!sun!guy
From: guy@sun.uucp (Guy Harris)
Newsgroups: net.emacs
Subject: Re: GNU emacs port to NCR Tower
Message-ID: <2370@sun.uucp>
Date: Tue, 2-Jul-85 00:10:17 EDT
Article-I.D.: sun.2370
Posted: Tue Jul  2 00:10:17 1985
Date-Received: Wed, 3-Jul-85 20:06:55 EDT
References: <2214@ncrcae.UUCP>
Distribution: net.emacs
Organization: Sun Microsystems, Inc.
Lines: 17

> However, my latest problem concerns the bit-fields used in
> the Lisp_Object. Neither the Tower (and Motorola PCC2 compiler)
> nor our SUN will allow signed integers in a bit field (the s.val part
> of a Lisp_Object).

Any C program that assumes that bit fields can be signed integers is broken;
K&R makes no such commitment - in fact, it explicitly mentions that the
PDP-11 implementation makes all bit fields unsigned.  Signed bit fields are
a nuisance to implement on machines without bit-field instructions; you can
extract a zero-extended field with a shift and an AND but you have to do a
bit more work to sign-extend a field.  Sounds like the GNU emacs code has
fallen victim to the "the whole world looks like a VAX" syndrome.

If anybody fixes this, it should be sent back to RMS to avoid further
annoyance.

	Guy Harris