Re: Take a Cup of tea : Part 2

[Marcus Brinkmann]

At Mon, 14 Aug 2006 17:51:55 +0200,
William DUCK <address@hidden> wrote:
> Hello Mr Brinkmann,
> 
> Sorry for the perturbation...
> 
> But I ask myself why the HURD project takes so much time to have a 1.0
> release ...

That's a very good question, and worth thinking about.  It provides
important lessons for any future or on-going effort.

One can't nail it down to a single reason.  You have to consider the
history of the free software movement and the history of operating
system research together.  However, of these two, I think that the
history of operating system research is more important.

In essence, it is an open research problem if a microkernel
multi-server operating system can succeed to provide a competitive
general-purpose operating system that scales up and down and is
deployed in a wide range of fields of endeavour.

Even if we limit the scope, the task is very challenging.  There has
been important progress in the last decade, but there are still some
(many?) question marks.

> Why did you choose coyotos that has only a commercial interest ?

Neal and I have looked at much of the relevant research in the last
years.  EROS/Coyotos is interesting because it tells us a lot about
how to structure secure capability-based operating design.  This has
lead to a much better understanding of the Hurd design itself, and how
to improve on it.

That Coyotos is also (not exclusively, as you portrait it) a
commercial project is true of almost every other interesting project
out there, and a clear benefit.

The question if Coyotos should be used to build an actual system
implementation is a question of utility, and depends only on technical
and legal matters, and little else.

> Is there nowadays a real alternative to GNU/Linux & Windows ?

It depends on what you mean by "real".  For specific applications,
GNU/Linux or Windows isn't even an option.  But the Hurd is certainly
targeted to compete with the systems you named, and in this context
the question is an important one, but one that has no known answer.
That's what we are trying to find out!

> You don't realize that you represented hope for many users in the
> world ...

I do realize it, and it is kinda pressuring.  But I don't have a magic
hat.  I have put a lot of effort into trying to understand the issues
involved, and I feel I am only about halfway to a solution.  That's
not meant to be discouraging, quite the opposite: Naivity has made
room for a path to understanding, which means that at least today I
understand much better what we don't know, but need to know to make
progress on these issues.

> The mailing-list is nearly empty, the l4-hurd cvs is frozen since many
> months, coyotos doesn't build ...
> 
> I have decided to ask the help to many experts ( bart massey, rene rebe,
> logic experts, HurdFR team... ) to build a fork.

You are welcomed to it, in fact, I can only strongly encourage you to
make your own experiences.  However, I have to ask you one thing:
Please do it in a non-disruptive manner.  There are very simple rules
of respect in human-to-human interaction that you seem to have
violated in the last days.  This is completely unnecessary, and
actually works against you.

Let me finish this mail on an up-beat: You have mentioned and
referenced interesting research projects.  For example, you have
raised the issue of what effect the choice of programming language has
on system design, among other things.  Well, sometimes we don't do
something simply because we are ignorant about it, so we reject it
because we don't have any idea what to do with it, and don't have the
time to learn at this point.  So, one thing you could do is to become
a domain expert on these issues, and report back to us with your
finding.  Be aware those that this does mean more than just sending
links and web page quotes, or asking others to do the work for you.
It means that you do the researching and studying, and that you get to
give the answers to the questions, combining knowledge from domain
experts in related fields to specific knowledge you add about the
Hurd.  This is harder than sending around emails, but it is also much
more rewarding, and it harms nobody!

For example, if you head over to
http://www.marcus-brinkmann.org/cap-papers/ you will find out that
there is not a single paper in there that ties operating system
research to language research.  That's a glaring gap in this
subjective and personal collection.  If someone where to collect the
most important papers about this topic, and would review them and put
them into the context of the Hurd project, that would be an important
contribution in my view.

Thanks,
Marcus