Re: Hurd/L4 as a class project

[Marcus Brinkmann]

Hi,

generally, if I am not mistaken, class projects have some of
constraints that are somewhat incompatible with free software
development (ie, you need some part of the system which you did for
yourself without help from others, whereas free software development
is often cooperative and incremental), but that shouldn't hold you
back.  It's just that those are issues you will have to cope with some
way or the other.  There is more than enough to do, you'll find
something interesting and useful I am sure.

At Mon, 7 Feb 2005 16:17:49 -0500 (EST),
Ivan Jager <address@hidden> wrote:
> with some OS hacking experience, but neither of us has hacked on L4 or
> Hurd before.

This means that a first priority for you will be to actually get some
experience with both, unless you want to stick strictly with the
driver framework, for which Hurd knowledge is of secondary
importance.

> We're considering something along the lines of adding an IDE driver,
> and possibly a network device driver,

As you correctly point out, there is no driver framework yet.

> and getting L4-Hurd to run off a
> filesystem on an IDE hard-drive, with the goal of being able to run more 
> useful programs. Right now I don't know how hard that is. Between the two 
> of us we'd be doing around 24 hours of development work a week,

Mmh.  If you expect to write a substantial contribution intended for
inclusion in the source, I think you can already mark some extra
shifts in your calendar :)

> I'm still catching up on the current status of the project so I don't
> know how much of user-land is working.

Not much.

> It sounds like it loads a console,

Actually, no.  There is just a very simple "console" in deva that, with
some hacks, is used for in- and output for testing.

> and can at least run banner, but I'm assuming that
> many syscalls are still unimplemented.

In fact, many syscalls are not syscalls :) But the equivalent to Unix
syscalls, yes, they are mostly missing.

> Also, If I understand
> correctly, the device driver interface layer (deva) is still being
> designed,

Is actually not being designed, as currently nobody is working on it.

> and the server model is under debate.

Not sure what that means, but the server model is actually the one
that is developed the furthest, and only implementation details are
being worked out, not over-all structure.
 
> So, firstly, I'm curious how people would react to a couple of newbes
> working on such a project, and contributing it back to the source
> tree.

We appreciate all help, and all interest in general.  Our own
resources are quite limited, and our task is huge, so you can not
expect much hand-holding as far as your class project is concerned
(Mathieu will hopefully agree, though, that we are able and
happy to hold hands if we march in lock-step).

> Secondly, I know that Mathieu Lemerre has been doing a fair
> amount of work on Deva. I'm wondering if it would be better to stay
> out of the way and try and work on something else, or if he could use
> the help. If we want to write device drivers, first we need an
> interface layer to write for.

From my point of view he was working on IPC issues, and only
implements one use of it in deva right now.  I don't know what he
wants to work on next, though, as the support for string items seems
to come to a first successful conclusion (I just need to review and
apply the patches).

I am not sure if we can assign you something and then promise to stay
out of your way until your class project is done.  If that is a
requirement for your class project, then that could be a problem.  If
you think this might be an issue, it may be useful to talk to your
teacher about the nature of free software development and how
important sharing knowledge and cooperation is.  If it helps, me or
Neal might be able to write something about your contributions at the
end of the project, if that would allow more cooperation during the
project.  All this assumes that you are serious about writing code for
inclusion (something that we will probably only find out much later).

Another issue: If you want to write code for inclusion, we need
copyright assignments to the FSF.  This might require a disclaimer
from your school.

You should also realize that this project is under rapid development
and many things, even quite fundamental design decisions, can change
from one day to another.  Depending on what you choose to work on, it
can get a bit rocky.

> Sorry if my understanding of the code is still a little flaky, but I'd
> like to feel out the project development process as soon as possible.

We are here, being feeled :)

Welcome to the Hurd, and happy hacking!

Thanks,
Marcus