On Mon, 13 Jan 2003 18:03:00 +0200
Erez Hadad <[EMAIL PROTECTED]> wrote:
> 1. OS's scope is a unification of the existing "Intro to OS" and "OSS".

Hmmm... Does it mean that a typical undergrad would now have only a
single semester of OS experience instead of two?

You may want to take into account that the "Intro" stuff (which was based
on Xinu) is simple enough to be taken on early semesters, while digging into
Linux kernel requires higher expretise in C and should be taken later
(e.g: after the student has done at least one software project).

What's the prerequisites for the new course?

> 2. The real-world OS of choice for the "OS" course is Linux. This means that 

"Mazel' Tov", better late than never (I still remember few years ago, some people
from the Technion boasted about the Solaris source licenses they will use....)

> 1. Opinions: Is Linux a good candidate for such a course, as opposed to, for 
> example, FreeBSD? Please provide good documentation sources, formal and 

Classic research literature is definitely BSD oriented and there are good
books (e.g: The Design and Implementation of BSD-4.3 by Sam Leffler,
TCP/IP Illustrated by Richard Stevens).

However, this situation is chaning rapidly. As an example I looked at the
conference proceedings for the FREENIX track of the Usenix annual technical
conference: while in 1999 BSD articles dominated the conference, in 2001
Linux and BSD has roughly the same number of papers.

Personaly, I would take Linux as the students would have higher chance to
get informal help regarding mundane issues (install, configure, tools etc.)

> informal, online and hard-copy, for both the kernel and API. Consistency 
> between the existing kernel and the documentation is a MAJOR factor (for 
> example, I heard many rumors about rapid under-documented changes in the 
> virtual memory).

A wild suggestion. Maybe use an older kernel (e.g: 2.2.x or even 2.0.x)
It is smaller and less complicated. In this sense, I liked the old "Intro"
course which introduced many important concepts at very early stage. The
problem was always with the next course ("Mivne") which was:
        - Not mandatory IIRC
        - Was given on proprietary Unices, so the only practice students
          got was writing "application code" (forks/pipes etc.) and not
          kernel code.

> 2. Tools: What is the safest and most comfortable way for students to hack the 
> kernel? VMWare? UML?

If they take this course before they did a major user-space project, they
are doomed anyway... Other than that, UML sounds more feasible (in modern
kernel) since anyone can use it at home as well.


Oron Peled                             Voice/Fax: +972-4-8228492
[EMAIL PROTECTED]                  http://www.actcom.co.il/~oron

"Normal people ... believe that if it ain't broke, don't fix it.
Engineers believe that if it ain't broke, it doesn't have enough
features ... yet." -- Scott Adams

Haifa Linux Club Mailing List (http://www.haifux.org)
To unsub send an empty message to [EMAIL PROTECTED]

Reply via email to