Re: debian-science and science-* packages
Hi Andreas,
On Mon, 2007-08-27 at 08:04 +0200, Andreas Tille wrote:
> On Sun, 26 Aug 2007, Christian Holm Christensen wrote:
>
> > I think a "Recommends" would be better than "Depends",
...
> See me previous comment about new cdd-dev tools. (They just turn
> Depends in the tasks files into Recommends.)
OK, I've missed that.
> > * Perhaps `-physics' should recommend `-statistics'.
> Well, we will have meta packages that will contain general scientific
> tools (statistics, typesetting and viewing in the current suggestion)
> which will probably be usefull in every specific science and that
> should be in the list of Recommends (or Suggests?).
For `-physics', both `-statistics' and `-mathematics' should be
recommended. The `-viewing' package is another story, and could
probably be suggested only. The `-publishing' set should probably also
be suggested (at least).
BTW, why is "gnuplot" in viewing? Although rudimentary, it does provide
a few tools to do statistical analysis, like least-square fitting, and
so on.
> > * The `-astronomy' package should also depend on what-ever
> > implementation that exists in Debian of `IDL' (Interactive Data
> > Language). For some odd reason, that language seems popular
> > among astrologists - sorry astronomers :-)
>
> I wonder whether it might be reasonable to implement a scheme like
>
> science-<x> and science-<x>-dev
>
> while the later contains developent libraries etc. In this case
> most probably IDL would go into the science-astronomy-dev dependency
> list.
I can only speak for High Energy Physics (both the particle and
heavy-ion "flavours"), but I think it applies to other sub-branches too.
In HEP we almost always need `-dev' stuff. It is very rare that people
get some sort of data and just feed it into a stand-alone program to do
their analysis. Most of the time, HEP-researchers (the ones that
actually do analysis) write rather long (and some times too long)
programs to do some weird stuff. After all, doing number-crunching on
10 variables/particle with ~4000 particles/event in some O(10000) events
is not something you want Mathematica to do :-)
I realise that we cannot cover all bases, and the `-dev' packages may
not be such a bad idea - I just wanted to let you in on a bit of
methodology.
> Thanks for your comments
No problem, and thank your for the great initiative.
Yours,
--
___ | Christian Holm Christensen
|_| | -------------------------------------------------------------
| | Address: Sankt Hansgade 23, 1. th. Phone: (+45) 35 35 96 91
_| DK-2200 Copenhagen N Cell: (+45) 24 61 85 91
_| Denmark Office: (+45) 353 25 404
____| Email: cholm@nbi.dk Web: www.nbi.dk/~cholm
| |
Reply to: