On Tue, 11 Mar 2008, Charles Plessy wrote:
Le Tue, Mar 11, 2008 at 12:26:57AM +0100, Christoph Berg a écrit :it appears to me that autodock and autogrid's Suggests on popularity-contest merely serve the purpose of pimping that package's popcon scores. Please remove the Suggests.Indeed, the rationale is explained by Steffen in README.Debian: ...
Besides the explanation given above I don't think that you can really
pimp a packages popcon scores that way. Popcon decides between packages
that are just installed and packages which content was really used by
a user of the system. So if some instances of popcon are installed on
boxes of users which installed it because of the suggests inside popcon
nothing really happens. It has just a minor influence on those boxes
where users really use auto* and would not have installed popcon if there
would not be the Suggests. Comparing the number of all popcon users
to those view users that fit this criterium the influence might be
smaller than a small epsilon.
On the other hand I personally do not like the Suggest because it is
just adding a dependency relation to a package that does not add any
value to the functionality of the package itself. The package would
work perfectly if you would replace the Suggests by a Conflicts. So
this is general missuse of Suggests and I would like it to be removed
for this reason - even if I think I have understand the problem which
is explained in README.Debian.
As a compromise I would be open for a discussion to make med-config
Suggests/Recommends popcon. This would lead to the effect that popcon
will be potentially installed on those machines that install auto*
via the med-bio package dependencies. The rationale behind this popcon
suggests would be that we get some information about the real usage
of our Dependencies of the meta packages and might be able to
adjust the Recommends / Suggests ratio according to these data.
What do you think about this?
Kind regards
Andreas.
--
http://fam-tille.de