Re: license packages
Michal Suchanek wrote:
> FWIW if the first alternative was always used what use for
> alternatives is there?
there's the following reproducable case:
1. you have neither iceweasel-l10n-de installed, nor any of
myspell-de-de, myspell-de-ch and myspell-de-at.
2. you have more packages installed that do recommend either
myspell-de-ch or myspell-de-at, and less packages than that
that do recommend myspell-de-de.
3. iceweasel-l10n-de recommends: myspell-de-de | myspell-de-at |
myspell-de-ch
4. since aptitude does install recommended packages by default,
you would imagine that an 'aptitude install iceweasel-l10n-de'
would result in iceweasel-1l0n and myspell-de-de being installed.
but that is not what happens. what happens is that aptitude tries
to be extra clever and will install iceweasel-l10n-de and either
myspell-de-at, or myspell-de-at plus myspell-de-ch, or myspell-de-ch
(depending on how the other packages recommendations are
distributed between the two 'unwanted' myspell packages).
here, aptitude is doing this *against* the recommendation that i set
as beeing the maintainer of iceweasel-l10n - it installs new packages
without respecting my expressed will in debian/control.
> If the input-all was always chosen it would make the alternative
> dependency useless, and the package would always work.
nope. per policy, the meaning of an alternative is (have not looked it
up, my wording follows here): "install the first package if none of the
alternatives is already installed". apt does implement it according to
policy, aptitude doesn't.
--
Address: Daniel Baumann, Burgunderstrasse 3, CH-4562 Biberist
Email: daniel.baumann@panthera-systems.net
Internet: http://people.panthera-systems.net/~daniel-baumann/
Reply to: