Re: 'mktemp' testing package dependency

[Sven Joachim <svenjoac@gmx.de>]

On 2009-06-17 15:23 +0200, Daniel Burrows wrote:

> On Wed, Jun 17, 2009 at 07:47:20AM +0200, Sven Joachim <svenjoac@gmx.de> was heard to say:
>> On 2009-06-17 05:59 +0200, Daniel Burrows wrote:
>> 
>> > On Tue, Jun 16, 2009 at 10:59:03PM +0200, Sven Joachim <svenjoac@gmx.de> was heard to say:
>> >> On 2009-06-16 21:44 +0200, Jason Filippou wrote:
>> >> > Downgrade the following packages:
>> >> > mktemp [1.6-4 (now) -> 1.5-9 (stable)]
>> >> >
>> >> > Score is 80
>> >> 
>> >> I wonder why aptitude is suggesting this, downgrading mktemp would not
>> >> really help.
>> >
>> >   According to packages.debian.org, mktemp version 1.6-4 doesn't have
>> > the problematic dependency, probably because it's a real package and
>> > not a transitional package.
>> 
>> Surely, but why does aptitude want to downgrade it to 1.5-9 then?
>
>   Sorry, I wrote that too late at night. :-)
>
>   I meant that version 1.5-9 doesn't have the problematic dependency.

Oh yes, that makes more sense.

>   He has a scheduled upgrade to version 7.4-2 from 1.6-4.  There are
> four ways to resolve the resulting dependency mismatch:
>
>   (1) Cancel the upgrade
>   (2) Remove mktemp
>   (3) Downgrade to 1.5-9
>   (4) Upgrade coreutils
>
>   Option (1) is knocked out because it would result in doing nothing at
> all, and option (2) is knocked out because it removes an Essential
> package.  That leaves (3) and (4), and they both scored equally
> according to the heuristic, so they showed up in arbitrary order.

Thanks, I understand now.  But I think my statement that this would not
really help is true anyway, because if you accept option (3) the next
full-upgrade will run into the same problem, this time the coreutils
upgrade being the only acceptable solution.

Sven