[Aptitude-devel] Bug#840902: install debtags; purge debtags leaves behind residue

[Manuel A. Fernandez Montecelo]

Control: tags -1 + wontfix
Control: close -1


Hi,

2016-10-16 00:36 積丹尼 Dan Jacobson:
>Package: aptitude
>Version: 0.8.3-1+b1
>Severity: wishlist
>
># aptitude install debtags
># aptitude purge debtags #but that leaves behind junk, so must then do:
># aptitude purge python3-debian python3-six python3-chardet python3-pkg-resources python3-apt
>using
>#cat /etc/apt/apt.conf.d/10jidanni
>APT::Default-Release "experimental";
>APT::Cache::AllVersions false;
>APT::Clean-Installed false;
>APT::Get::Fix-Missing true;
>APT::Get::Purge true;
>APT::Install-Recommends false;
>Binary::apt::APT::Keep-Downloaded-Packages true;
>APT::Keep-Downloaded-Packages true;
>Aptitude::CmdLine::Always-Prompt true;
>Aptitude::Purge-Unused true;
>Aptitude::CmdLine::Show-Why true;
>Aptitude::CmdLine::Show-Deps true;
>Aptitude::CmdLine::Verbose 1;
>Acquire::http::No-Cache true;
>Acquire::PDiffs true;

I think that we discussed this many times already... debtags depends on
python3-debian and python3-apt for example, which in turn might be
recommended by other packages, depending on the state of the system.

If a recommended package is installed by any means, it doesn't get
auto-removed -- aptitude doesn't know if it was installed on purpose or
not, or if the user started to depend on it since it was installed, so
conservatively it's not removed.  Otherwise, the removal of any package
could trigger removals of any other packages kept only for being
recommended packages -- and many people rely on this behaviour to keep
packages installed untill all rev-depends that recommend it are removed.

So this is not going to be changed, it's a characteristic (or undesired
behaviour, depending on how one sees it) of Debian and the system of
dependencies.

Another reason can be mishandled auto-flags, in which case there are
lots of bugs about this already.

If you can provide more info which is not any of this cases please
reopen, but in the meantime I'm closing it because it's almost surely a
dupe of one of the cases above.


Cheers.
-- 
Manuel A. Fernandez Montecelo <manuel.montezelo at gmail.com>