Kurosu a écrit : > Matthieu Fertré a écrit : > >> So, in my opinion, the detection of duplicated rectangles must be done >> in Map::OptimizeCache(). Methods Map::ToRedrawOnScreen() and >> Map::ToRedrawOnMap() should stay silly methods. >> > > I think that on the other hand, as long as Map::ToRedrawOnWhatever is > not in a time-critical path, it should do the job: that's the best spot > to optimize. Let me show you my train of thought about this. > > Prerequisite: > Have something to actually benchmark what is faster (I have my own idea > about this). > > Steps: > - Following Kyle's notice, having duplicates is not such a big problem: > (a) contains (a), so (a) can probably be removed during the occlusion test > - Is sorting the list then performing occlusion the fastest, ie is > sorting really needed? >
No but if you don't sort, you have to iterate over all the items. I suppose that the efficiency depends on the size of the lists. > - If that is faster, then isn't it faster although to have an > automatically sorted list by adding each element at its correctly > ordered location; you have a smaller set to work on > Well probably right, and in that case, you can remote it directly at insertion if it's a useless rectangle. Anyway, OptimizeCache is perhaps totally useless, do we really gain something with it ? I'm not sure there's so many duplicates (but it is true that the SDL blit function costs a lot!) Regards, Matt (gentildemon) _______________________________________________ Wormux-dev mailing list Wormux-dev@gna.org https://mail.gna.org/listinfo/wormux-dev
