On Wed, 2011-11-02 at 09:31 +0530, Rahul Sundaram wrote:
On 11/02/2011 07:54 AM, Kevin Kofler wrote:
> Stephen Gallagher wrote:
>> * #683 - Zif as default PackageKit backend for desktop users -
>> (sgallagh, 17:03:32)
>> * AGREED: ZifByDefaultForDesktop is refused as a Feature for Fedora 17
>> (sgallagh, 17:07:32)
> IMHO refusing this was a bad idea when the more general rule was originally
> decided and is still a bad idea now. You're essentially blackmailing zif
> upstream: "Either you fully support command-line users or you will never be
> the default in Fedora's PackageKit."
I don't think this is the argument. Having potentially different
behaviour between command line dep resolver and gui dep resolver is very
problematic and this is a important concern.
Exactly, Kevin is skewing the reason of the decision here. The reject is
not based on the whole backend as is but purely on the depsolver
algorithm. As soon as yum and zif share the depsolver (not just the
transaction depsolver in rpm, but the depsolver solving which packages
to put into the transaction), then zif would be allowed as the default
backend for PackageKit. There were some proposals/ideas by the rpm
developers to implement this depsolver in librpm although I do not know
whether there was any progress in this recently.
No matter how far down the wrong road you've gone, turn back.