[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Some Hurdy pointers needed.

From: Marco Gerards
Subject: Re: Some Hurdy pointers needed.
Date: Tue, 29 Mar 2005 21:26:10 +0200
User-agent: Gnus/5.1007 (Gnus v5.10.7) Emacs/21.3 (gnu/linux)

Michael Han <> writes:

> 1. Is there anything that is closer to the user-level that would
> clearly separate microkernel based OS (i.e. Hurd) from monolithic ones
> (i.e. Linux)? From user p.o.v., it's rather hard to come up with clear
> advantages of switching to it at this point--that is putting aside the
> fact that it's still under development stage. If translators are fully
> exploited (i.e. use of XSLT, implementation of DocBook standards,
> etc.) that may be a different story. And I may be talking over my head
> if I'm overestimating the scope of translators.

In the Hurd users get more power without compromising the security.
For example it is possible to write and run some system services as a
regular user, for example filesystems.  You can have no or multiple
UIDs, etc.

> 2. Intuitively, one can tell that having an extra layer below an OS
> creates an overhead to the performance. (I know last sentence may
> charge some people with emotions, but I'm only speaking generically
> with the general theoretical difference between monolithic kernel and
> microkernel in mind.) Is there a work under way to create
> microkernel-based processor at the hardware level? Or has the
> processors evolved in such way it already gives monolithic ones the
> type of performance advantage that microkernel may seem to enjoy?

You could have a look at L4.  But I think microkernel based OS'es will
remain a bit slower, although you will have more flexibility.
Hopefully someone will proof me wrong.

> 4. Continuing from my question #1, is there any active development
> under way to implement XSLT-type of programs as translators?

What would you have in mind?  Both XSLT and translators are
interesting IMHO, but I don't see how to combine them.  Perhaps you
could make some kind of "symlink" and auto process all XML files or
so, but that seems silly to me. ;)

Please read a bit about translators on  They
don't really translate in the way you think they do, as in
transformations of data like XSLT does.  They are just servers that
can be accessed via the filesystem.  For example a filesystem sitting
on some node, a device node or even a server that just speaks a very
specific protocol (password server, console server, etc).


reply via email to

[Prev in Thread] Current Thread [Next in Thread]