> >In regard to kernel design, I would propose that we adopt an MS-Mach > >(multiserver, not Microsoft) like approach, albeit with kernel space > >"servers" as kernel addons. Thus, we could have an fs module that > >handles file desrciptors, etc, a posix module, etc., etc., etc. This > >would accomplish two things: > > To some degree, this is what BeOS already did/does. > BFS is a file system add-on. Many things were/are. > In fact, you can add almost anything, as I understand it, > to the BeOS kernel by making a kernel add-on. No, no, no, you misunderstood completely. I, of course, want filesystems as addons, but I also want the code that loads the filesystem addons to be an addon. Nor can you add anything to the kernel; your efforts in this regard are severely circumscribed. For instance, you can't write an addon that intercepts syscalls, which would, of course, be required for a posix addon. > >2. It vastly increases the future flexibility of the kernel. > > I don't know that much about MS-Mach - why not mail me offlist > and we can chat about it. Sure, all right. -Nathan -- Fortune Cookie Says: "If you understand what you're doing, you're not learning anything." -- A. L.