On Dec 10, 2007 7:05 AM, Brian Salter-Duke <b_duke@bigpond.net.au> wrote:
1. The ability to display molecular models that can be manipulated by the
reader. This function is available in an open source application called
Jmol. A php extension is available to run in mediawiki and Jmol does run
on, for example, the Jmol Wiki. I have tried to get this added to
Wikipedia, where of course it would also be useful, but the technical
folks, think the php is not sufficiently free of security concerns. I do
not know php but I am trying to get this issue solved. Unfortunately the
author of the Jmol php extension is too busy to do this.

If you can provide us with further details, I can get it started on the sandbox wiki we have running on the server.
 
2. The real innovation of our materials is that they allow the user to
prepare data for a variety of different computational chemistry programs
and then run the data, either directly in real time, getting the results
back on a web page, or by queueing the job if it is larger and getting
the results back later on a web page or by e-mail. Our materials use a
variety of quite complex CGI scripts that handle the materials from html
forms and menus and then run the code. While queueing the jobs is
necessary in some cases, this is really only to restrict access to the
disk for storing large temporary files. The CGI scripts prevent the user
from entering large jobs. In this context "large job" equates to "large
molecule" with the cpu time rising with some some power of the molecular
size. For teaching, small molecules only are needed and these take little
time. When a job is queued, for example, it leaves a URL link on the
page returned by the CGI script. The user can keep clicking on it until
it returns more than a header saying try again later.

While currently we use both open source and commercial codes, the
commercial codes can be replaced by open source ones by writing a few
more CGI scripts.

This use clearly needs a server that can run these CGI scripts and the
programs they call. It does not have to be the server that handles the
forms page. For a while we ran one application 3000 km away from the
main server.

I've got a 40kg server with triple power supplies, dual CPUs, triple hard drives and dual ear muffs to be able to hear with it all that we might be able to use :) Let me know if this sounds okay and I'll get a flavor of Linux up and running on it.
 
I have no idea what the equivalent to CGI scripts is within mediawiki
and how we can run forms (the quiz extension may be a start here), but
I'm willing to learn. Whether this is feasable, I do not really know.
However, I think that what I am suggesting could have wider use.
Wikiversity needs to have pages that are dynamic in ways that wikipedia
does not need. Web based learning has to be more than reading text and
looking at pictures on a computer screen. I think this is the greatest
challenge for the technical people on wikiversity.

One of the main purposes of the sandbox server was to allow us to break out of the restrictions of MediaWiki. Forms in PHP are dead easy. Even if you don't have a programming background you could learn in half an hour. MediaWiki can have forms, but it requires some highly complex code and a few hacks here and there. We would be developing real applications to facilitate our learning systems, as opposed to this reliance on MediaWiki for what clearly warrants a dedicated application.
 
There has been a suggestion that these kinds of computer calculations
can be handled by distributed computing using users own PCs, rather like
the SETI project. These could be usefull for very large jobs in some
cases on WV, but I do not think they are suitable for my applications
where we just need a few seconds (maybe up to 100 secs) on a linux
machine. Most of the codes we use are unix based. Distributing computing
is not required and it would delay getting the response to the learner.

If it's a few seconds of calculations, why do we need a computer to do it in the first place? Can't students run the algorithms on their own machines? In any case, for a few seconds of calculation the overhead of a distributed task system like SETI is totally out of the question, a server would be best.
 
I would welcome suggestions for dealing with this.

Brian.

Let me know what you think of these.

Draicone