Date: Sat, 13 Jul 1996 20:13:39 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: dennis@etinc.com (Dennis) Cc: terry@lambert.org, hackers@freebsd.org Subject: Re: Kernel Config (Was: GENERIC Kernel Debate) Message-ID: <199607140313.UAA05909@phaeton.artisoft.com> In-Reply-To: <199607131607.MAA16924@etinc.com> from "Dennis" at Jul 13, 96 12:07:37 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> >> [etc, etc.] > >> > >> at 7 bucks a meg...who cares? > > > >People who distribute boot floppies? > > This is silly. There's no reason to put a kitchen-sink kernel on > a boot floppy. Having a couple of different boot floppy kernels is > easy. The "generic kernel" issue is one of compatibility, content > and size. It's an issue of "what hoops must I jump through to run FreeBSD". It's *not* a content issue, unless content must be static. If content must be static, I claim the design to be flawed, since this is how I defined it. If you dislike my definition, provide a better one; your dislike have been noted, but is not persuasive in any way without an alternate definition having been provided. > >People who want to sell their sync serial cards which require > >proprietary drivers into the "I couln't build a kernel to save > >my life" market. > > Hardly. Anyone who thinks that someone is going to be compelled to > use unix simply because he doesnt have to build a kernel has his > head, neck and shoulders buried in the sand. I do *not* believe they will be compelled to use UNIX. On the contrary, I believe they will be compelled by usability issues to use NT or some other OS other than UNIX. I dislike this idea, however, I have seen much in the way of supporting evidence, and nothing to the contrary from the heated opposition to the idea. > Based on our experience with LINUX loadable modules, building a > FreeBSD kernel is a lot easier and requires less support than loadable > modules. Unless you implement a perfectly seemless loadable module > interface, they are arguably more difficult and more problematic than > static kernel modules. This is *precisely* the model I keep proposing. It is *precisely* the model Windows NT uses, and Windows NT has a provably superior dynamic linking technology on which they implement their model. > My message was in response to your goals, which seemed to be mostly > addressing the memory footprint (ie, discarding initialization code, etc). My goals are to have the bare minimum necessary kernel, no less. And no more. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199607140313.UAA05909>