Yet another bug. This time filters vs selections

[Boudewijn Rempt]

On Sunday 13 September 2009, Dmitry Kazakov wrote:

> Hmm.. It seems to work with lightweight filters (like Invert) only...

I can't say I'm too surprised, really. About the performance hit of iterating 
over selections: you get that hit in any case when you bitblt, since that 
opens two iterators, too. Last sprint, in Berlin, we discussed ways of 
optimizing iterators, and we concluded that iterators should cache way more 
aggressively, i.e., right now we retrieve the next tile on every tile boundary 
hit. This means that in a worst case scenario we look for the next tile 63 
times more often than really necessary.

I'm not sure whether that's still true with your new tile engine, though.

-- 
Boudewijn Rempt | http://www.valdyas.org