From owner-freebsd-hackers Sun Apr 20 06:44:59 1997 Return-Path: Received: (from root@localhost) by freefall.freebsd.org (8.8.5/8.8.5) id GAA21437 for hackers-outgoing; Sun, 20 Apr 1997 06:44:59 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id GAA21426 for ; Sun, 20 Apr 1997 06:44:52 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.4/8.6.9) id IAA00954; Sun, 20 Apr 1997 08:44:43 -0500 (EST) From: "John S. Dyson" Message-Id: <199704201344.IAA00954@dyson.iquest.net> Subject: Re: Price of FreeBSD (was On Holy Wars...) In-Reply-To: <335A00EF.E5A@wgold.demon.co.uk> from James Mansion at "Apr 20, 97 12:41:35 pm" To: james@wgold.demon.co.uk (James Mansion) Date: Sun, 20 Apr 1997 08:44:43 -0500 (EST) Cc: freebsd-hackers@freebsd.org Reply-To: dyson@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > John S. Dyson wrote: > > Well -- I don't want an OS like NT. Microsoft has already done that, > > and it shows *interesting* performance charactistics. Of course, I > > admit that it would be very nice if we could maintain more consistant > > interfaces, but frankly, I have used the latest version of Unixware MP > > code also, and I don't like that either... It just doesn't work nicely. > > > > John > > I don't see why one would equate NT's performance characteristics with > an improved level of modularity. For a start, you wouldn't need to have > the same LRPC/'microkernel' (ha!) approach. > I don't know if my response was publically posted or only privately. Bottom line is that I don't think that our system is mature enough yet to stick ourselves with a legacy design. There will be a time when things will be better defined -- but not yet. Frankly you NEED a HAL for NT or DDK specs for other commerical U**Xs because they won't give you the source code for free (or a reasonable price), and not having the specs would simply stop development. With only a very small investment, you can read all of our source code, develop your own idea as to how to do things and do them. I had the UNIX source code about the time when the specs for writing device drivers came out for it. I thought it was a joke mostly because it was very incomplete, and writing device drivers using just that public document would limit users to less that what was being used in the existing code. (The newer docs are only a little better.) I don't know about you, but previously messing with U**X internals and looking at 386BSD or even Linux for the first time, I could do the things necessary to write a driver. To me, it is almost all the same. If you are wanting enough to do a port (like a full HAL), sorry -- but right now I don't think that it is a good idea to constrain the architecture. Until we do a few non-X86 ports, I don't want to freeze anything, or religiously adhere to something that is wrong. Someday we will have a better idea of what is needed. (I know what is needed now, but I don't have time to document it, and it will likely change soon.) So, unless you are a guru, don't do a port to a new arch. If you are sharp, you can probably do one. (I don't think that FreeBSD plans to support more than 4-5 archs at any given time (note that our estimate is going up, for good reason :-)).) John