> I think you're greatly over-complicating this, which is just going to > lead to weird behaviour when things turn on and off, and arguments about > which level some painting should be in. Mhhh maybe, but why not think at an architecture for med-to-long term, and implement only most common cases (like all as today, and none) ? And maybe the medium level at the moment as all without animations and transitions ? To not change existing base classes maybe also this could be done only inside a skin, and see later if refactor/move ...
> Before "features" like this go in, I'd like to see some example of a > real problem caused by any of these. OK, no real problem, but in many Enterprises you know that many PCs are outdated or have graphics cards with poor performances, so why not provide a way to disable all the visual stuff that can cause slowness ? And there are also Companies with Polices that doesn't want applications with too much graphics effects ... while at the opposite for Consumer Users this is wanted. A Test implementation could be simple (ok, nothing is simple), and we could decide later if keep or extend, or remove ... Many of my ideas come from my direct experience on Enterprise Companies, where Rich Client Applications are not wanted and Browser + AJAX + sometimes Flash are the only solution ... so trying to push technologies like Pivot there is difficult for many reasons, so my trouble to optimize Pivot for this type of users. > The gradients in Pivot are software-rendered, so that's not an issue. > Java doesn't currently offer any way (short of using JOGL) to get > hardware accelerated gradients. >From Java 6 Update 10 this has changed, the Graphics pipeline has been optimized a lot for most common operations like this, at least on Windows. >> But the best could be to have a utility class to query the graphics >> engine to tell is the desired effect is Hardware-enabled (but I don't >> know if this is possible and how to do, sorry), maybe in a way OK, I don't know if this is currently possible, and also if it is maybe it's too much complex. I could try to make some experiment on this, for the 1.2 release ... Comments ? Thanks, Sandro
