As shown in this thread, parallel branches run their children in the "wrong" order, contrary to the rest of the tree. It happens on the 2nd run and remains like that. This is done intentionally:
Hm.. I'd rather give interruption priority to the early branches. It seems more logical and follows the general rule, from left to right. I find reversed order of parallels quite questionable and misleading solution. Even the icon suggests the direction, may be it should look like this?
It turns out that If you want uniform order in your tree, you need to place all tasks under the parallel branches in reverse. Or you can write custom classes for every parallel task. And still, on the first iteration, the order will be reversed. To be honest, none of these solutions I like.
@Justin, maybe you could implement this mechanics natively? Add a checkbox to the tree settings, in which direction to run parallel branches. Make it unchecked by default, nothing will conflict with earlier versions. It seems easier for you to inverse execution list in the source, than for us to copy-code each parallel class and hack currentChildIndex..
And please, for mercy's sake, make it consistent between the frames.
Hm.. I'd rather give interruption priority to the early branches. It seems more logical and follows the general rule, from left to right. I find reversed order of parallels quite questionable and misleading solution. Even the icon suggests the direction, may be it should look like this?
It turns out that If you want uniform order in your tree, you need to place all tasks under the parallel branches in reverse. Or you can write custom classes for every parallel task. And still, on the first iteration, the order will be reversed. To be honest, none of these solutions I like.
@Justin, maybe you could implement this mechanics natively? Add a checkbox to the tree settings, in which direction to run parallel branches. Make it unchecked by default, nothing will conflict with earlier versions. It seems easier for you to inverse execution list in the source, than for us to copy-code each parallel class and hack currentChildIndex..
And please, for mercy's sake, make it consistent between the frames.
Last edited:
