Re: Hurd/L4 as a class project

[Doug Jones]

Matthew T. Brewer wrote:


> This particular class' purpose is to get students to make significant
> contributions to open source OS projects under licenses like the GPL, 
> or the BSD license. So we can certainly work on GPL'd code, and
> contribute it back. Collaboration with the current developers of the
> project is strong encouraged, so we really don't need exclusive access
> to some chunk of the source code. It's our problem as students to
> explain what it is we've actually done.
>
> Any interesting project has unsolved problems and thus somewhat
> changeable interfaces etc. It's no fun writing code someone else has
> written before. My questions come more out of interest in efficiency,
> and avoiding work duplication, just like any other developers working
> on this project. We're not asking for hand holding, merely trying to
> avoid doing work someone else is already doing, and avoid writing code
> that someone else will have to rewrite later, because it wasn't what
> the l4-Hurd development group as a whole wanted. Writing a bunch of
> code that will not be accepted back into the source tree would be a
> purely academic venture, which (ironically enough) is not the goal.

[snip]


Once again mbrewer and Ivan have asked the Frequently Asked Question: 
What can we do to help?

Seems to me we are seeing this question more and more frequently, and 
the answer remains the same:  There is no easy answer.  The number of 
people hacking the Hurd is small, and I think many here will agree that 
their limited time is better spent coding rather than trying to explain 
everything to those of us who are sitting near the bottom of a really 
big learning curve.  The Hurd is simply not the most accessible of projects.

But of course that must inevitably change.  How long will it be before X 
is running fairly well, and the beginnings of a driver framework are in 
place?  How long before people start wading through the thousands of 
Debian packages, trying to see which ones will run on the Hurd on L4? 
These things might seem far away now, but the day will come, and such 
days tend to come with disturbing suddenness.

I'm sure there are lots of us lurking on these lists who do not possess 
the skillsets, experience or time needed to deep dive into Hurd hacking, 
but who have nonetheless grokked the potential of this project.  I have 
never compiled any sort of kernel, and I'm not favorably inclined to 
mucking about with my partitions on a regular basis, but I'm still 
looking for some kind of meaningful contribution I can make.

There may be those who don't wish for the Hurd to become too accessible 
too fast, simply because of the increased noise levels it would create 
on the lists.  But I'm sure there are lots of people who are far more 
competent than I, and who could make very significant contributions, but 
who are staying away because of the daunting height of those barriers.

What was the one thing that first made Linux accessible to me?  Knoppix. 
 A live-CD.

IIRC the idea of a Hurd live-CD has been discussed before.  It's been 
sitting on a back burner for a while, so I'm taking a moment to stir it 
again.

It is suddenly much easier now to make a live-CD.  Lots of people are 
doing this now, and there's plenty of writing on the subject.  In 
addition to the Knoppix Hacks book, there's

  http://ibuild.livecd.net/

and

  http://www.babytux.org/articles/howto/how2livecd.php

and probably lots more that I haven't stumbled across yet.

Maybe it's time to take another look at all this, and try to determine 
what would be involved in getting something like a linux-qemu-l4-Hurd 
stack running on a live-CD.  Would be excellent if we could have scripts 
to rebuild the ISO from sources periodically, and automatically push a 
Torrent out on an RSS feed (hey I can dream, can't I...).

There may be relatively few programs that can be run on the Hurd/L4 
right now, but that number will presumably increase, and hopefully that 
number will eventually take off.  It would be very handy if the live-CD 
was already available on that day, so that people like me could start 
testing programs in that environment without having to climb such a high 
learning curve.  (And, it might even benefit actual Hurd hackers who 
frequently admit that they don't even have a working Hurd at the 
moment...  ;-)

mbrewer and Ivan may not be interested in doing something like this, but 
perhaps somebody else is.  I imagine there are lots of people out there 
who would find deep Hurd hacking rather intimidating, but who could 
tackle a live-CD project with relative ease.  And I suspect a lot of 
lurkers here are eagerly awaiting a chance to taste the delicious meal 
that Marcus and colleagues are serving.