McStas ILL Web site trial

[Kristian Nielsen]

> Date: Wed, 01 Dec 1999 16:01:35 +0100
> From: Farhi <farhi at ill.fr>

> I've corrected the bad link. Yes you can announce the site, but I'm not sure yet of its
> location (will this stay on the ill/tas group zone ?). Perhaps it's better to wait the
> Dec 10th meeting, where I will raise the question.

Ok, I will wait then.

> The number of components is not growing quite fast. The main problem is that physicists
> are very bad programmers, and the source code is usually a black box, with no help,
> etc...

Yes, this is clearly a problem to be addressed. One idea that came up
was to have some sort of journal for components, where you would submit
a component with a paper describing the algorithm for peer review. If
the component and paper was accepted, you would then get scientific
credit, just as for an ordinary paper. But it may not be so easy to set
up in practice.

> I think you're right : we should maintain a unique set of components, shared on both web
> sites. You told me you know how to update those files (mirror). I never tried this
> before.

Well, I think I would have no problem setting up a little program that
would automatically copy pages from the remote to the local server. This
could use a web proxy to get through the firewall (I assume this is what
your Netscape does).

The problem is that we really want both of us to be able to add entries
to a component list in a single file mirrored on both servers (Risø and
ILL). I cannot think of any good solutions right now, maybe we can wait
a month or two and then do something reasonable?

> Also, can McStas search for components inside sub-directories in it's lib path ? This
> would enable to sort components by class (sources, monitors, etc)

Yes, I could add that to the next version. I think I would prefer not to
search arbitrary subdirectories, since this is potentially very time
consiming and is also not portable (to Windows/Mac). How about having a
fixed set of subdirectories (for example "samples", "monitors",
"sources", "optics", "misc") that are searched?

> Thomas Hansen has noticed a McStas limit concerning the number of instrument parameters.
> He uses in it's D20 simulation about 20 to 30 input params !! As mcstas did not work

Hm, I do not understand this. There IS a limit, which I will remove when
the need arises, but that is set to 1000 parameters. I just tried an
instrument with 299 input parameters, and that worked fine. Do you think
you or Thomas could give some more details (maybe an error message)?

 - Kristian.