Andy Tanenbaum has released 3.1.2-B2 of MINIX, a preview version of 3.1.2. We reported extensively on the initial release of MINIX3 last year. The most important new feauture of this release is the inclusion of the X windowing system.
Andy Tanenbaum has released 3.1.2-B2 of MINIX, a preview version of 3.1.2. We reported extensively on the initial release of MINIX3 last year. The most important new feauture of this release is the inclusion of the X windowing system.
I guess the only wm that can run on it is twm?
Well, seeing as that’s more than enough for some people… I’m sure it’ll do fine.
I’ve kinda been waiting to see X ported; now the work on porting graphical applications to Minix can really take off…
By the way, does anyone know what version of X was ported? Xorg? XFree86? Kdrive/something else?
Xorg 6.8.2
I guess the only wm that can run on it is twm?
It’s got POSIX, X11, and gcc. That probably means you can compile several other window managers.
(Well, at least the kernel-mode kernel, if that sentence even makes sense.)
I think that’s cool, it makes the code easier to maintain, which, for an OS, is important. I hope MINIX 3 will delier on it’s promise of stability. I was, however, a bit disappointed reading that it requires 8 MB RAM. Slightly on the fat side, for an embeddable OS, in my opinion. I remember, back in the days I worked with netWare 3 (3.11, 3.12), I was told the kernel was really tiny, some 2-3 thousands of lines of code – and assembler, at that. Then you had the NLM – netware loadable modules, which would “plug in” into the kernel. NetWare 3.11 would work just fine with 4 MB RAM, happily serving files to dozens of users.
(Well, at least the kernel-mode kernel, if that sentence even makes sense.)
4000 lines in the micro-kernel. Really, though, the layer above (server-layer, which includes the file system, memory management, etc.) contains more of the code and is just as important. So I don’t think the number 4000 is particularly relevant for maintenance. It is usually good for stability to limit the number of lines of code that have highest privelege, though.
Edited 2006-03-29 11:51
> So I don’t think the number 4000 is particularly
> relevant for maintenance. It is usually good for
> stability to limit the number of lines of code that
> have highest privelege, though.
It is important because a microkernel system enforces strong modularity (although modularity can also come without a microkernel). There is a cleanly defined interface between the kernel and server processes (hope this is the case for the new Minix. It’s a general concept of microkernels but the old Minix didn’t abide to it very well). With such an interface, maintenance is mainly concerned with the maximum number of lines per module, not total number of lines.
There are some neat aspects to this. The way device drivers are handled sounds good and will make their development massively easier than most other operating systems. They will certainly need to expand the range of device drivers and making the process dramatically easier is an excellent portent.