theflashbum / bitmapscroller Goto Github PK

In sampleBitmapData, Rectangle clone() call can be inlined. If BitmapData size doesn't change, no need to create a new instance each time as it results in lots of garbage collection/memory allocation in an oscillating manner. Just remember to lock the BitmapData prior to setting pixels during the sample() loop. If one needs to account for different BitmapData dimensions for differnet orientations/states, consider creating a fixed size to account for all possible sizes, or ~2 set sizes.

Instead of recursely calling sample() which involves calling the costly calculateCollectionStartIndex() method, consider the possibility of using a while loop and an incrementing index. Another way is to precalculating the start index once (outside the sample() function.) and increment the index on each iteration.

Furthermore, inline whenever possible. Ever considered using Haxe for inlining and the possibility of writing directly to memory through Alchemy ? (even though that's more suited for per-pixel drawing/vector buffer). This could improve the performance by a certain degree..

Scroll-wise, a drag toss effect might be better as the scrolling is too fast now given the mouse-to-screen/totalScrollWidth ratio. Elastic snapping to the edges of the nearest indexed item would be a great touch.

I'm using BitmapScroller and I love it. But with the rescaling of the browserwindow I want to redraw all the bitmaps to the appropriate size (full-window-size) so I'm setting the 'bitmapDataCollection = <NEW_ARRAY_WITH_BITMAPS>'. This works fine except for the rectangles staying the same sizes.

Proposal:
In the function 'indexCollection', before the loop, the collectionRects-array need to be reset:

collectionRects = [];