Re: one or two modest proposals.

[mooney@oxygen.aps1.anl.gov (Tim Mooney)]

Mitch Nelson writes:
(By the way, Mitch is referring primarily to software for x-ray
diffraction here, although the ideas are generally applicable.)

> So here is the combined proposition:
> 
> To use tcl/tk for our purposes, we  really only need to supply a handful
> of  compiled code  modules.  The  macro-language  in tcl  seems to offer
> everything  else that  one would  need to  develop an  attractive set of
> commands, widgets, menus, etc.

Those code modules might also get dragged into, say, IDL.  I wouldn't
want to get hung up on this possibility now, but if otherwise arbitrary
decisions regarding the call interface would make general use of the
modules easier or harder, I would vote to make general use easier even
at some expense in elegance, performance, or programmer time.
Something to keep in the back of one's mind, perhaps?

> Both the  tasks,  of  developing these  few  compiled code  modules, and
> designing the command  set and gui's, are  activities that could be done
> in an efficient and timely way by us as a community.

I think this is a great idea.  We'll pitch in.

> So, I'm willing to send anybody who would like to join either or both of
> these efforts, a  prototypic command set,  and/or a list of the routines
> that we think we need.

Please send both as they become available.

> Perhaps Robert Fleming will be willing to coordinate the command set and
> gui  discussion.  And John  and I could  keep  track of the  code module
> specifications and who's working on which code modules.

We may need a location and source-code control for shared header files.

> The ground rules  would be that  all codes are  written in ansi-c and/or
> tcl/tk and that all sources are given for free use but not for re-sale.

ok

> Some other  considerations that  we've developed are  that we would like
> the result to be transportable, and that we would like to use the 'igor'
> interface style for gui's where feasible and appropriate.
> 
> 'Transportable' here means,  layered on EPICS, rather than imbedding one
> in the other  in an  inextricable way. By  this  approach, we hope to be
> able to implement the system off-line or at our home institutions, using
> alternative hardware interface packages.

One hard part here, I think, is deciding on the optimal place to make
the cut.  I suggest we take the time to get a consensus on this point.

> 'Igor' is adapted as a  user-interface style  because most of the people
> that we've talked to seem to like it.

Yup.

> A  conservative  estimate  of the  compiled  code   development seems to
> indicate a total  investment of just a few  person-months. John seems to
> feel that the tcl/tk  codes are much less work  than even that. Deciding
> on the command  set, given our  starting point, will  probably take us a
> few weeks.

I think this is worth at least a man year, particularly since much of the
work is certain to be reuseable in other x-ray subfields.

> Regards to all,
> M. C. Nelson
> 

Tim