[LON-CAPA-dev] lonc/lond, construction space, etc

Gerd Kortemeyer lon-capa-dev@mail.lon-capa.org
Sun, 26 Jan 2003 21:29:33 -0500

[Moved from users to dev]


Guy Albertelli II wrote:

> > Ron Fox wrote:

> > >   The export problem is perhaps a specific instance of a more
> general
> > >  problem that could be solved: Allow access to construction space
> > >  from any server, not just the home login server.
> >
> > That would be extremely cool!
> I think I would argue against this.

My  two cents would be that only the web interface of the constuction
space is going to work cross-server, while the construction space
continues to resides on one and only one server. If the author wants to
sftp, mount, samba, or appleshare the disk (direct access), they still
have to connect to their home server.

I agree with Guy, trying to distribute sftp and other direct access
methods would create all of the horrible mess he describes. When doing
direct access, however, it would be great if the permissions and
ownership could somehow be figured out. Users continue to mess up the
permissions ... not sure how this can be solved, it is hard ...

In summary: lonc/lond would be the file transport layer for the web
interface only.

BTW, what is all this WebDAV business (besides creating weird error
messages on our servers)? I am currently in New Orleans at the Educause
NLII Conference and actually just registered for their "Next Generation
Course Management System" Focus group in Tuscon, AZ, on March 7th
(anybody wants to come along?). Anyway, on the "wishlist" for the "Next
Generations CMS" is WebDAV authoring support - can we do that against
our construction space?

Next thing is that in the lonc/lond rewrite we probably should look at
the OKI stuff right from the start, so that at least we do not do
anything to keep it from working. Check out http://web.mit.edu/oki/ and
in particular http://www.lon-capa.org/oki.html, which is the outcome of
a technical discussion with the folks over there. Note that OKI lives
both above and below lonc/lond, which all of us found a clean solution
to get around the problem that OKI has no concept whatsoever of a
multi-server peer-to-peer network. For us, that means we need no
revolution to do OKI, but should keep it in the back of our heads as we
write our stuff.

... well, this was probably more than two cents. Of course, in all of
these revolutions, we need to make very sure that we have a very very
stable production environment in Fall 2003, which probably means that a
lot (if not all) of the stuff above simply has to wait. In talking with
the other MSU folks who are here in New Orleans the biggest issue they
have against LON-CAPA is bugs, meaning, they doubt that it can serve all
of MSU because it is not a production but a research system - we need to
remove that obstacle and perception. Bugs are the number one reason that
we are having problems gathering central support. So: let's put all of
these cool features on the back burner, unless we can be very very sure
that they work and do not destroy other functionality.

- Gerd.