home *** CD-ROM | disk | FTP | other *** search
Wrap
Received: from sloth.swcp.com (sloth.swcp.com [198.59.115.25]) by nacm.com (8.6.10/8.6.9) with ESMTP id LAA07444 for <executor@nacm.com>; Fri, 23 Jun 1995 11:39:33 -0700 Received: from iclone.UUCP (uucp@localhost) by sloth.swcp.com (8.6.9/8.6.9) with UUCP id MAA12353; Fri, 23 Jun 1995 12:38:58 -0600 Received: by mailhost (nextstep Smail3.1.29.0 #11) id m0sPDc9-000YbmC; Fri, 23 Jun 95 12:38 MDT Message-Id: <m0sPDc9-000YbmC@mailhost> Date: Fri, 23 Jun 95 12:38 MDT From: ctm@ardi.com (Clifford T. Matthews) To: Mark Adler <Mark.Adler@quest.jpl.nasa.gov> Cc: executor@nacm.com Subject: Re: nextstep In-Reply-To: <9506231749.AA01445@quest.jpl.nasa.gov> References: <199506230147.SAA24616@nacm.com> <9506231749.AA01445@quest.jpl.nasa.gov> Sender: owner-paper@nacm.com Precedence: bulk >>>>> "Mark" == Mark Adler <Mark.Adler@quest.jpl.nasa.gov> writes: >>> NEXTSTEP port Mark> Oh thank you, thank you. Mark> (To the person who wondered if NEXTSTEP was some sort of i/o Mark> support, no. NEXTSTEP is the only decent operating system, Mark> window application environment, and development environment Mark> in the universe. It is the birthplace of Executor and the Mark> port of NEXTSTEP to non-680x0 architectures was the Mark> inspiration for the synthetic cpu in Executor. Cliff can Mark> correct me if I'm wrong here.) In my opinion NEXTSTEP is in fact the best operating system/windowing system currently available. The first 10,000 lines of ROMlib, the precursor to Executor, were written on a 1 MB Macintosh Plus with no hard drive, only one floppy drive and no printer. I used "Lightspeed C" version 1 (the precursor to Think C) and had to use five different floppies for the source since there wasn't too much room left on an 800k floppy with Lightspeed C on it. The filesystem was actually written on a PDP-11 (no, I'm not making this up) running UNIX. Executor itself, was actually born on a Sun3/60 taking over the frame buffer directly. A port to X-Windows on the Sun3/60 followed, but yes, NEXTSTEP was where Executor was first commercially released. NeXT's port of NEXTSTEP to non-680x0 architecture was not the inspiration for the synthetic CPU in Executor. I've been familiar with portable code and assembly language since before NeXT existed. I knew from the start that if we were to port Executor to a non-68k environment we'd need a synthetic CPU. Originally I designed the synthetic CPU when I was in bed with pneumonia. I hired Mat Hostetter to implement my design -- I thought I would do the design work and hold his hand as he implemented it. HAH! Mat came up with a much better design and needed none of my help to implement it (although I had to do some revamping of Executor's internals so that it would mate nicely with Mat's synthetic CPU and also not be freaked out by the different byte ordering involved). Although we like NEXTSTEP quite a lot, we shifted all our development to Linux because NEXTSTEP licenses are expensive, NeXT doesn't fix their kernel bugs and it's a pain to compile Free Software Foundation tools under NEXTSTEP right out of the box -- we get a lot of leverage out of the FSF tools. Since we made the shift we also found that we could use hardware watchpoints under gdb under Linux (a big win) and lately we've also been using a Linux only debugging tool called "checker". It was sad to move our development to Linux, but NeXT didn't seem to care and in the long run it's helped us. For people doing high level development the NEXTSTEP development environment probably can't be beat right now (but that's changing). However, writing a Mac emulator isn't a high level undertaking. Furthermore, as an Independent Software Vendor, we need a market for our products and being only able to sell to NEXTSTEP users greatly narrows the potential customer base. --Cliff ctm@ardi.com