delta wrote:xalign stops animating and is fixed at 0.5, the rest continues. Not that hard.
I'm not sure I like that, though. We have code that appears to be updating two properties, but with your semantics, sometimes will be updating only one property. That seems kinda weird to me.
What I think is going on here is that I'm thinking of ATL as a script that runs and updates properties, while you're thinking of transform properties as being largely independent, with ATL being a means of varying those properties over time. (Am I right?)
There's also the problem that we have multiple non-independent properties. For example, the xpos property can be updated when the pos, xalign, align, angle, and radius properties are set. This is good and necessary, as these properties are not independent of each other... but it does mean the interactions could become very complex.
That makes me want to shy away of a model in which the animation of some, but not all, transform properties are overridden.
I'd be inclined to agree with you more if you wouldn't specifically tout holding over the properties themselves as a feature (which it is... except when it isn't). So the ATL blocks are already not atomic.
We have a model in which one ATL block runs at a time, per image tag. It's the properties that are held over, not the running ATL.
I just don't think of these in terms of "blocks" at all, but to consider each property and its animation just as a more refined version of the property with half a dimension added.
I don't believe this is the right model to have.
In general, the question here is this one:
Is it logical desirable behavior that clicking too fast or skip mode stops ATL blocks?
And the answer is no.
Well, the real problem is not "clicking too fast" or "skip mode", it's "running a show statement". And I think it's fair that running a show statement that changes what is being shown should also change the ATL block associated with what is being shown.
Showing a second image early will always be a bit of an edge case. It's not clear to me that there is behavior that is better than the current one (which requires the game-maker to resolve this edge case). I don't believe that the current proposals are better than the current behavior.
I am wondering if there needs to be a variant of the show statement that changes the image being shown while keeping around the ATL state, and if that would help with most of this.
we couldn't start a transition at the current position.
I meant to say "transform" instead of "transition".