[Qt4-preview-feedback] Qt 4.5-rc1: solved a performance problem with model + filter, introduced a new one?

[Adeodato Simó]

* Olivier Goffart [Friday, February 13, 2009 10:49]: 

Hello, Olivier, thanks for your reply.

> On Thursday 12 February 2009 12:26:34 Adeodato Simó wrote:

>> 1. The old problem
>> ==================
> [...]

> This was indeed a known problem which has been fixed.

Great! :-)

>> 2. The new problem
>> ==================

>> Can somebody help me figure out what's going on?
 
> It's difficult to say anything without any testcase.
 
> Many speed optimisations have been done.   But it is assumed that the call
> to the data index and rowcount are relatively fast.
 
> If you can give me a example i can use to reproduce, I'll be able to
> profile it.

Okay. This is a Python program, but I guess you'll be interested in a C++
test case. I'll see about providing one during the weekend. (If Python is
okay, by all means let me know. ;-)

All my index/data/rowCount calls should be O(1). Well, they use
QModelIndex::column(), ::isValid(), ::internalPointer(), and
QAbstractItemModel::hasIndex() and ::createIndex(), but I assume all those
are fair game.

Thanks,

-- 
Adeodato Simó