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