Re: Am I missing something?

[Eric Fluger]

OK. Here's the sequence, step by step as any-random-user would 
do it:

go to google:
http://www.google.com/

type in: gnu hurd l4 <return>
which takes you to:
http://www.google.com/search?hl=en&ie=ISO-8859-1&q=gnu+hurd+l4

The first item listed is titled: Hurd on L4
If you click on it you go to:
http://www.nongnu.org/l4hurd/

This site presents a bold opening statement:
This project is dead and the information here is outdated. Look 
at http://savannah.gnu.org/cgi-bin/viewcvs/hurd/hurd-l4/ for 
more up-to-date information.

The url is a live link. Click on it and you go to:
http://savannah.gnu.org/cgi-bin/viewcvs/hurd/hurd-l4/

And YES! after going through every single directory (almost) I 
did find one chang that was 3 weeks old and another that was 8 
days old. (On my first visit I just tried the first few 
directories, which mostly have not been touched for at least 
several months.)

So I guess there's something going on, but one sure has to be 
committed to find out about it.

It's also worth noting that the next three listings returned by 
google are:

savannah.gnu.org/projects/l4hurd/
https://savannah.gnu.org/projects/hurd/
kerneltrap.org/news/hurd

All of which contain old news. It would be nice to get an update 
onto kerneltrap. Perhaps the really issue here has less to do 
with what is or is not happening, than how the project looks to 
the average data-mook.

I think it would be helpful to the general public standing of 
the gnu/hurd project to make the L4 porting effort more visible 
if you can do so without raising unreasonable expectations.
(I suspect that some folks who were excited about the project 
when it started now figure that they have linux or *bsd and that 
seems to be working OK so why bother with hurd?)

The prospect of an L4 port is enticing in four ways:  :

-- Speed: L4 is tauted as being faster than mach.  Performance 
is important to some for its own sake.  It is important to 
others as a portability factor.  (No sense in porting to my 
installed hardware base if my installed base to slow to run it.)

-- The usually cited advantages of a true microkernel 
architecture.

-- The prospect of something new: I suspect a lot of folks have 
"been there, done that" attitude toward any *Xish environment 
running on top of mach.  The prospect of a shiny new kernel 
might get their attention. (Personally, I find the multi-server 
architecture pretty interesting no matter what it runs on, but 
I'm not sure how many of my colleagues in the work-a-day world 
would agree.)

-- Obsolescence prevention:  The very fact that the hurd can run 
on more than one kernel is a big plus. hurdaganda :-) usually 
stresses the modularity of the multi-server environment. Servers 
can be individually added, removed, updated without effecting 
the others. However, the whole system seems tied to mach. 
(perhaps the opposite of MK/L4-Linux?) This leaves the kernel as 
prospective point of obsolescence. If you can change ANY 
component, even the kernel, in a way that is transparent to user 
processes you can upgrade the system incrementally far into the 
future (perhaps to the point where it becomes almost 
unrecognizable) no matter what kind of changes in hardware 
architecture, network architecture or user expectations come down 
the pike. (no pun intended). That should keep people interested.

Hope some of this has been constructive.

Eric Fluger
201-860-9608 Desk
201-892-7669 Mobile

On Wed, 11 Aug 2004, Alfred M. Szmidt wrote:

>   The site does link back to a repository on Savannah, but the change
>   log shows the last change to be last April.
>
> Which repository are you talking about?  And what site?
>
> By the way, could you for the future specify what exactly you are
> talking about? (i.e. don't use "the site links", but "http://foo.bar
> links").  It is awefully confusing, and it wastes time since we must
> ask you what item you are really talking about.
>
> The hurd-l4 module in the Hurd CVS repository was touched last week by
> Marcus Brinkmann...
>
>   > Doubt it, GNU Mach is around 700Mb gziped.
>                                     ^^
> Kb! Not megs...