[Debichem-devel] Reorganizing debichem tasks (Was: Sponsor shelxle upload)

[Andreas Tille]

[Now also switched to your better subject]

On Thu, Feb 09, 2012 at 11:59:52AM +0100, Michael Banck wrote:
> > 
> > We have invented bio-cloud which contains a *subset* of the packages
> > inside the bio task which does not need any graphical interface and thus
> > is dedicated for installations in clouds, clusters whatever.  When
> > designing a task there is no reason to put a package exclusively into
> > one single task.  Tasks are not exclusive categories but finally these
> > should match user needs.  So if there are potential users who might have
> > an interest in non-graphical applications it makes sense to provide such
> > a task file.
> 
> Good idea, so we could have debichem-base have non-graphical things and
> a debichem-workstation task which pulls in the main graphical apps
> (pymol, avogadro, gnome-chemistry-utils)

That's perfectly in the line what I intended to suggest.

BTW, you can depend from other meta-packages like

  Package: debichem-workstation
  Depends: debichem-base

I even considered "resolving" this kind of Dependencies in the tasks
pages.  For this reason science-biology has

  Meta-Depends: svn://svn.debian.org/blends/projects/med/trunk/debian-med/tasks/bio

and

  Meta-Suggests: svn://svn.debian.org/blends/projects/med/trunk/debian-med/tasks/bio-dev

However, this feature is not implemented (for years) and the entry is
ignored.  The reason why I did ot yet imlemented this is that I plan to
inject tasks into UDD which would simplify things a lot and specifically
this feature might come nearly "for free" while otherwise it is some
effort to implement.  So please do not trust on this feature on the
tasks pages for the near future - but the meta-package dependencies are
resolved for sure.
  
> > So in short I would group the development tools according their topic
> > and not according to their programming language.  I repeat:  Feel free
> > to derive from this suggestion if you think the typical chemist will
> > think differently.
> 
> Well, the big difference IMHO with python is that you can use it ad-hoc
> in the interpreter, just pulling in modules, for doing quick analysis or
> computations.  The C/C++/Fortran libraries are really only useful if
> you are writing a complete application.

Finally it is your decision.  If you consider this difference important
enough to derive from the principle we were following that's perfectly
OK. 

Kind regards

         Andreas.

-- 
http://fam-tille.de