Hi, > IMO, it is risky to leave the ADG's renderers in this state. Fair enough and I'm sure you know more about the ADG innards than me.
> But the question is, what about some third-party who extended the renderer > and added another style > not in this list? They could override the setStyle method? The fix in question I had in several ADGs in production for several years and not run into any issues (including navigation key support), but they are not using any RTL ADGs. > 3) Changing this styleProp "if" to a lookup on some mx_internal > object/map. Rating: -0.1. Again this leaves the flag set to true, but at > least gives someone a way to hack in other properties (plus the check will > be much faster). Don't think speed is a huge issue here when changing layoutDirection. How often would you change the layoutDirection in a real application? > 4) Change AdvancedDataGridItemRenderer to always register for validation > when invalidateDisplayList() is called. Rating: 0.5. This will make the > renderer conform to the lifecycle "spec", but will simply result in > validateNow() being called on the renderer, which you were trying to > eliminate. The fix doesn't try to remove it altogether just stop it from being called multiple times when it not required. The speed improvement is noticeable. The initial issue is that it gets called multiple times per cell when scrolling/moving about the ADG. There are other issues with entire rows changes when they shouldn't but that's enough issue. There's some code in my whiteboard that demonstrates this. I have a look at the options in more detail once I get some time - likely to be end of the week/next week. Thanks, Justin
