Hi.

On Thu, Nov 20, 2014 at 06:25:13PM -0800, Charles Thorley wrote:
> I find Minix3 to be extremely interesting, and attractive (at least in
> principle).

The advantages of Minix are purely theoretical. All the different 
servers can be restarted when they crash, but that does not make the 
cause for the crash go away. I saw this happening a few times when I 
worked with Minix. A server crashes, it is restarted, it crashes again 
because the cause for the crash did not magically go away, it is 
restarted… You get the idea.
In many situations this does not even work. If the filesystem server, or 
whatever this thing is called crashed it _should_ be restarted, but this 
not possible, because the necessary binary for the restart is 
unavailable, because the filesystem can't be reached.
Most parts of the environment do not even work correctly. The last thing 
I remember was something like

        find /usr -name \*foo\*

which crashed miserably, because find can only recurse through 
filesystem structures with no more than 256 entries or so.

In my personal opinion Minix is no foundation to build on. There is a 
reason why Linux Torvalds with his 'hobby project' got much more thrust 
from the community than Andrew. S. Tanenbaum with his professional work.

Kind regards,
-Alex

Reply via email to