Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/3/84; site delftcc.UUCP Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!think!harvard!seismo!cmcl2!philabs!rdin!delftcc!sam From: sam@delftcc.UUCP (Samuel C. Kendall) Newsgroups: net.micro.att Subject: Cc with automatic -f is not enough Message-ID: <130@delftcc.UUCP> Date: Thu, 27-Jun-85 18:17:09 EDT Article-I.D.: delftcc.130 Posted: Thu Jun 27 18:17:09 1985 Date-Received: Sun, 30-Jun-85 03:09:09 EDT References: <104@gwsd.UUCP> <476@gitpyr.UUCP> <339@cuae2.UUCP> <199@hadron.UUCP> Organization: Delft Consulting Corp., New York Lines: 19 > ... [I]f the 3B2 n e e d s the '-f' flag, it > should automagically be included at compile time. > > Joe Yao hadron!jsdy@seismo.{ARPA,UUCP} That's a start, but what about object code portability? Currently, a program compiled without the -f flag on systems with floating point won't run on systems without fp; conversely, a program compiled with -f will be too large, and run much too slowly, on a system with fp. (Or will the interpreter be ignored at runtime on systems with fp?) Automatic -f won't help this problem. What is needed is for the interpreter to be dynamically loaded on systems without fp -- ideally at the time the first fp instruction is encountered, otherwise at load time. Sam Kendall UUCP: {allegra,cmcl2,ihnp4}!delftcc!sam Delft Consulting Corp. ARPA: delftcc!sam@NYU-CMCL2.ARPA Phone: (212) 243-8700 USMail: 432 Park Av S, New York, NY 10016