On Sun, Feb 19, 2012 at 11:27 PM, David Arno <da...@davidarno.org> wrote:
> When I looked into the idea of us simply porting Flex to haXe to solve > the JavaScript-target issues, what you describe was one of the killer > issues that made haXe a no-go. > > haXe doesn't do anything clever to target the DOM and > JavaScript-specific component libraries, thus if we followed the haXe > approach, we'd end up with Flex trying to render via canvas elements. > The canvas is far too slow in all browsers at present for this to be an > option. > > Therefore I think with Flex our only choice is to do the complete > opposite of haXe and to write JavaScript-specific renderers to replace > all the skin classes in Flex. That is the only way we can hope to get > good performance. > That's a bit quick and blunt as an assumption - canvas is pretty fast in modern desktop browsers (I should know, as I maintain a canvas-based flash api for haxe), by the time you have a _stable_ javascript backend for flex, this won't just be for desktops but pretty much standard for mobile hardware. Though, in all honesty, I think it's best to rely on standard HTML and CSS markup for layout, that's what it was designed for. I don't see how having "javascript specific renderers" is any different from writing it in another language that is compiled down to the host language - how these are to provide better performance eludes me. - Niel > David. > > >
