[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[sdpd] Re-using software pieces
Lachlan wrote :
>I have been told verbally that both Armel and
>myself are on the new IUCr Crystallographic Computing
>Commission (with David Watkin of Oxford University (author
>of the Crystals Single Crystal Software) as Chair
>but as I have not received anything on letter-head,
>The following are just personal comments.
>
>At the Computing Commission meeting in Glasgow, David
>Watkin considered it very important to create a
>publically accessible collection of both commissioned
>and donated datasets relevant to crystallography,
>(single crystal, powder, etc) so software developers
>have the best chance of creating crystallographic
>software that is as robust as possible. (this would
>also benefit for users who wished to use the data)
>Obviously this would be a long term and continuous
>effort.
Trying to maintain a minimum of discussion at the SDPD
mailing list :
I would add that the Computing Commission could (should...)
create a bank of well documented source codes. Those codes
should be preferably deposited under some GNU licence or
equivalent.
Re-inventing the wheel will not always produce one running
more true.
The last code I produced (ESPOIR) is a patchwork of old
Rietveld method + Struplo + RMCA codes + diffusion factors
data taken in Fullprof + etc.
If a good subroutine is able to translate the space group
Hermann Mauguin notation into symmetry operators, it should
be in the source code bank, clearly.
At the origin, the CCP14 distributed only executables.
Fortunately, on can find some sources here and there now.
The biggest piece of sources is in Fortran code inside the
CCSL (Cambridge Crystallography Subroutine Library -
at ILL - http://193.49.43.3/dif/ccsl/welcome.html) but this
is not enough. There is almost nothing available in Visual C++
code for instance, which would show how to implement
a good GUI in MS Windows (the Rasmol code was
available, coming from industry).
I have heard about the future availability of the new Fullprof
code, rewritten in pieces of reusable subroutines. This is certainly
good news.
So many programmers had previously voluntarily made their
code complex, that trying to reuse them is waste of time. The
European Software Institute seems to be convinced about the
usefulness of reusing and organizes workshop on that question :
http://www.esi.es/Events/Year98/ERW98/moreinfo.html
I do not see why this should only concern industry and not
academic research.
Nevertheless, mentalities have to be changed before a real community
(similar to the Open Code Linux community of developers) of
crystallography computing people emerges in the academic research world.
Considering the amazing cost increase of many commercial software and
databases, such a real community would certainly benefit to all.
My opinion about all commercial crystallography databases is
that they are the exact counter example of what should have been
done : free access to data freely given and part of a humanity
patrimony. The fault belongs to International Institutions which
should have make pressure enough for obtaining International funding
rather than individual/laboratories fees.
The future PubMed free access of medical publications online
http://www.nih.gov/welcome/director/pubmedcentral/pubmedcentral.htm
is easily understood in the context of the increase of research forces
in order to solve more quickly human health problems. The high
cost of scientific publishing brakes peoples in poor countries.
Crystallography plays a non-negligible role in life science.
Armel Le Bail
http://www.cristal.org/
PS - PowBase includes now 113 experimental powder patterns,
of which exactly 0 spontaneous donation ;-)
http://www.cristal.org/powbase/