Ashley, on r126 I'm seeing this - well the settings flyouts closing on a click anywhere on the flyout. If I interpret the thread correctly, it should be ok in that build, right?

I'm trying to come up with a generic approach, namely an event sheet that can be included in just about any Construct 2 game layout to provide Win8 snapped behavior with a minimum of additional work. I'm hoping someone can vet the approach below, since there are some unknowns (to me) with underlying implementation in Construct 2.

Right now I have set up an event sheet that I include in all the other layouts, and the event sheet looks like the following:

<img src="http://jimoneil.blob.core.windows.net/share/win8events.png" border="0" />

It seems to work, but wondering about a few things:

1. I need to know each of the layer names or at least how many layers are in a layout to generalize this. Right now I assume a "GameLayer" on each, some could have others or not use the naming convention. Is there a way to enumerate the layers in a layout?

2. I'm relying on Groups to deactivate responses to interaction when in snapped view (discovered that hiding the layer and setting time scale to 0 wasn't enough :)). Each layout now has a Group named "UnsnappedMode" - is it ok to reuse that name across layouts? The included event sheet then refers to "UnsnappedLayer" and gets the correct one from the context of the current layout.

3. I'm deactivating a group while within a group, e.g. group "SnappedMode" deactivates itself when transitioning out of snapped view. Is that ok.

4. Lastly, the transition to and from snapped mode isn't as fluid as I'd like. I'm using Letterbox Scale mode, but when going from snapped -> full landscape, you can see what looks like a transition to scaled first (IOW, I see the game screen in top left of the confines of the snapped view) for an instant before it goes to full on landscape. The reverse is true as well, seems like the game screen is cropped before the snapped view kicks in. I've tried flipping which layer I make visible/invisible first, but the flicker is still there.

I'd like to write a function (that's included in multiple layouts) that will hide all but two layers of the current layout. Each layout may include differing numbers of layers so I'm having trouble figuring out how to loop through them.

If I could toggle visibility of all layers at once that would work too.

My app (http://bit.ly/X9BmLv) is failing certification because of startup time (which for Windows 8 needs to be < 5 seconds). It runs just fine (and starts up much more quickly than that on a Surface RT).

In order to prune the startup time, what are my options? I don't have a lot of sprites in the game, though two sprites come with 26 animation frames each.

1. Are all of the images for all of my layers loaded as part of the initialization process? If so, is there a way to defer? My first screen doesn't use any of those 26 images.

2. Is there a rough threshold on how "big" my app can be and still pass the startup time constraint on Win 8? I read the performance tips blog post, but that's more about how quickly the app runs, which is not my issue.

3. Workaround for this in Win8 is an extended splash screen, so I'm going to give that a shot, hoping I can tweak the c2runtime.js (if needed) to pull that in.

Thanks for the offer of looking at the code! Capx is here: bit.ly/X9BmLv (it's my first project so if I'm doing anything "bad" would love to know).

There's some randomness to the whole thing, so on the Game layout, if you just let it run without moving the guy, you'll likely see a phantom collision within a minute or so.

I have a game modified off of the simple Vertical Space Shooter template. The 'enemy' sprites are generated every second, and I have a animation frame of 26 images, for which I assign one at random when the new sprite is created. (Each sprite represents an object starting with one of the letters of the alphabet).

I have another sprite that moves around to collect these items, and collision detection seems to work fine... except

there appear to be sprites generated which don't appear visually but do trigger the collision detection. I captured some of the properties to the screen in other fields and they look like legitimate objects, they just aren't displaying (as if opacity were 0 - but it's not).

I even see one triggered way before it's logically even generated - since I have a 8 second delay (for reading a message screen) before I start the object generation, but a collision is detected pretty much as soon as the layout starts.

Any ideas on how to track this down?