Date: 08 Apr 2001 23:38:21 +0200 From: Dag-Erling Smorgrav <des@ofug.org> To: Dale Chulhan - Home <dchulhan@uwi.tt> Cc: "chat@FreeBSD.ORG" <chat@FreeBSD.ORG> Subject: Re: Follow up: More Win vs NIX Message-ID: <xzpn19rqdgy.fsf@flood.ping.uio.no> In-Reply-To: <3AD053AF.4743DEC4@uwi.tt> References: <3AD053AF.4743DEC4@uwi.tt>
next in thread | previous in thread | raw e-mail | index | archive | help
> I emphasize the point by quoting the previous author: > "Unix does not use a monolithic kernel. Most Unix implementations > do,....." > <- does this guy know what he is really saying?? > He goes further by saying: "there is no reason why it can't run > on a > microkernel"<- again showing that it does not. Hrm. I wrote this. To clarify: There is not a single entity that you can call Unix. There is a collection of operating systems from different vendors which were derived from, or written to behave like, AT&T's original Unix operating system, some of which use a monolithic kernel and some of which don't. Thus, the statement "Unix uses a monolithic kernel" is meaningless. Of those operating systems we choose to call Unix, at least two that I know of use a microkernel: GNU/HURD and OS X (or rather, Darwin). In addition, there is a newly-started effort to write a Mach-based microkernel called xMach that will run a BSD (probably FreeBSD) userland. Note that FreeBSD itself does use a monolithic kernel, although most commonly-used drivers are dynamically loadable. And yes, I do believe I know a bit about this subject - I've been a FreeBSD developer for three years now, give or take a couple of weeks, and a Unix user and admin for several more. How many years' experience do *you* have writing operating systems? > Yes, I admit that you don't have to reboot for SOME driver > installations (you can compile and load drivers at runtime), but the > fact remains that the fundamental problem of a bloated kernel and of > having to LOAD drivers and services into ring 0 still exists. In FreeBSD 4.x and 5.x, most loadable drivers will be automatically loaded when you try to configure the devices they drive - e.g. typing 'ifconfig fxp0 inet 192.168.0.1' will load the fxp driver if it isn't already loaded or compiled into the kernel. > 4. Yes, many RFCs were developed on the UNIX system. However > Kerberos was not ported as the author says, it was written according > to RFC. In fact most UNIX guys complain about Microsoft's > non-standard implementation of Kerberos because they used 5 reserved > bytes in the protocol.... Meaning they violated the standard, which says these bytes are reserved for future expansion *of the standard*, which is not the same as saying they're reserved for whatever purpose the vendor chooses. > 6. I stand corrected on the IP Address change. I always rebooted > after I run ifconfig. I was not aware that reloading the interface > after running ifconfig on Linux would avoid having to reboot. There is no such thing as "reloading the interface". Just running ifconfig with the appropriate parameters immediately changes the IP address of an interface - or adds an IP address to an interface, if for some reason you want an interface to respond to several IP addresses. > However, according to Sun Documentation: http://docs.sun.com, the > reccomended procedure involves running sys-unconfig and rebooting. That's because every time a Sun user mistypes something or doesn't realize that changing the IP address will interrupt established connections, and calls Sun to complain that It's Not Working, Sun loses money. This is precisely the same reason why Windows used to claim that you needed to reboot when you'd changed the IP address (though if you just clicked "Cancel" it would work). Us Open Source types can afford to tell the user "tough cookies", but Sun and Microsoft can't. DES -- Dag-Erling Smorgrav - des@ofug.org To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?xzpn19rqdgy.fsf>