[Yum-devel] [PATCHES] Repository with fixes and new operations test frame work

[Florian Festi]

Florian Festi wrote:
> seth vidal wrote:
>> okay - I'm looking at these patches (thanks to jbowes for the git help)
>> and I'm still seeing FakeRpmSack and the stub installed() in there. I
>> thought we decided that doing it that way was not all that helpful and
>> it would be better to implement it as exists() in PackageSackBase and
>> get around that?
> 
> I stick with that because it was the less invasive way of getting things 
> up and running. exists() is a be bit generic IMHO. But if we want to 
> move that way I can provide a patch if no one else wants to do it 
> (should be quite trivial). On the other hand I see no reason why that 
> should block anything.

Ok, rebased the patch set to current HEAD (don't try this at home -  §$%! 
3-way merge). There is now a patch implementing PackageSackBase.contains() 
as replacement of installed(). Sorry, I really couldn't make friend with 
"exists" [*] (Although I am in a sufficient existentialist mood right now). 
Feel free to use query-replace... :)=

Florian

[*] The reason why i chose "contains" over "exists" is that "exists" 
questions at a global level while the method is restricted to the sack 
itself. "exists" should also not be affected by adding/removing to/from the 
sack - but the implementation is (and should). So the name needs to express 
the fact that the pkg needs to be part of that sack and "contains" does that 
more clearly.
In fact it is even worse as "exists" doesn't even make clear if the 
existence of pkgs or the sack itself is in question.