-dev metapackages

[Frederic Lehobey <Frederic@Lehobey.net>]

Hi,

Andreas Tille <tillea@rki.de> (2008-05-09 15:32:32) :
> On Fri, 9 May 2008, Manuel Prinz wrote:

> >I'll have a closer look but at the first glace it looks sensible. I just
> >do not see a good reason what the reasoning behind the mathematics-dev
> >task is.
> 
> The rationale behind the science-mathematics-dev package is that
> we do not want to flood users hard disk with lots of development stuff
> if he only wants to use some mathematical applications.  So this
> is a practical reason for the meta packages.  There is no such
> reason for the repository and I don't think it makes sense to
> stick to this for the VCS hierarchy.  That's what I meant when I
> wrote you should feel free to change things you don't feel reasonable.
> 
> >Or why other tasks do not have a corresponding section.
> 
> Also practical metapackage reasons:  There are just no so much other
> -dev packages so they do not need a separate meta package.  But IMHO this
> meta package considerations are void for the VSC hierarchy (in contrary
> it might be even a bad idea to implement this 1:1) because we need
> a structure according sciences to attract the right experts.  The
> developers who are working with the repository are expected to know
> themselves which package is development related and which not.

  Fully supported.

  One of the goal of these tasks packages is to create live media from
them. So I would roughly add (it is a very untrue statement in the
details): if a software has a GUI (and/or a menu entry) then it goes
to 'field', if it has none (libraries...) then it goes to 'field-dev'.

Best regards,
Frédéric