KSortFilterProxyModel

[Stephen Kelly]

John Tapsell wrote:

> 2009/11/29 Christoph Feck <christoph at maxiom.de>:
>> On Monday 30 November 2009 00:18:46 Stephen Kelly wrote:
>>> I haven't seen any support for the idea of moving it out of kdelibs
>>> despite the incompleteness. So unless someone else tries the testcase
>>> and agrees that the class is incomplete I'll just let it slide in too if
>>> that's the consensus.
>>
>> We are not in API freeze yet, so you can still change new API when this
>> is required to fix bugs. If you think your class is not ready to be made
>> public for 4.4, you can disable it for now until 4.5 (by not installing
>> its headers and documentation). This is inconvenient for any developer
>> that already wants to depend on the class, but better than maintaining a
>> "broken by design" class forever.
> 
> +1.  I like Stephen's proposal of replacing KSortFilterProxyModel, if
> we can find someone to spend a few minutes moving things about etc.
> Everytime I do something like this, I break kdelibs, so I'd prefer
> someone experienced did it :-D
> 
> I'd like to find a bit of time to look over the new version - in
> particular to benchmark it.  It seems to be very easy to end up with
> an O(n^2) filter or worse.
> 
> John

Yes, I have some ideas on improving the KRecursiveFilterProxyModel. My 
current implementation is broken in some ways. I should have it working and 
unit tests started by the end of the day.

I will move/copy KSortFitlerProxyModel this morning into the relevant places 
in kdebase.

All the best,

Steve.