[Aptitude-devel] Bug#699191: Bug#699191: aptitude don't depend on apt, but don't work correctly without apt

[Manuel A. Fernandez Montecelo]

Sorry, I somehow missed this reply...


2015-11-07 23:39 Axel Beckert:
>Hi,
>
>Manuel A. Fernandez Montecelo wrote:
>> >This has been discussed just recently on the Aptitude mailing list:
>> >
>> >http://lists.alioth.debian.org/pipermail/aptitude-devel/2012-December/003251.html
>> >
>> >It's currently not a real-life issue, because apt considers itself
>> >Essential (even if it isn't) and hence both, apt-get and aptitude will
>> >not remove apt unless they're convinced that you know what you're
>> >doing.
>>
>> Should we perhaps add at the packaging level an extra depends on apt?
>>
>> Even if there could be other ways to address it, this looks to me the
>> more straightforward one, and I don't imagine that having the extra
>> depends will hurt, unless apt changes dramatically in the future.
>
>Indeed.
>
>> And very few people will want apt removed from the system even if
>> aptitude would not require it.
>
>Still, if someone's really picky, this could count as RC-level bug.
>:-)

I don't understand... What could count as RC-level bug, this report?

I don't know if there is a better way to solve it, really.  The solution
should be in libapt, providers of downloading methods and the dependency
system.  But since the problem reaches aptitude in this case, we have
limited ways to deal with this.

We could require the apt package hardcoded in aptitude's resolution, but
that would be hardcoding it with the same effects as doing it at the
packaging level, only that it would be more difficult to modify and undo
if things change (if they move from "apt" to "libapt-download-methods",
for example).

So IMO we have the choice to address it in this way, adding extra
dependency from aptitude on apt at the packaging level, or leave open --
I prefer the former :)


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