Weird CocoonJS results - Screens & Capx provided

Hi again, I'm working on my run & jump project again and targeting android phones with CocoonJS. I stripped my .capx down to the point it can be readable by anyone.

This is the related .capx

I have been experiencing the following for the last few releases and decided to post about it now. I don't know if it can be fixed by changing something in the capx or it's a bug but I got too far into development (talking about the full capx) and I can't change the logic too much after this point, somehow hoping this be a bug :p

In short, this is how it looks like on desktop browsers. The way it's intended:

However my samsung galaxy s2 displays it like this:

<img src="https://dl.dropbox.com/u/3138530/2012-07-09%2012.56.58.png" border="0">

Notice the gaps between the tiled white sprite on the back. More importantly, the sprite on the right side of the platform can not follow the object it's attached to in a predictable way. Its distance from the platform changes with each instance.

Here is another:

<img src="https://dl.dropbox.com/u/3138530/2012-07-09%2012.56.23.png" border="0">

This is the CocoonJS zip you can test on your android device:

This is probably CocoonJS related instead of Construct but there's a chance it's not. Thanks for reading.

Edit:

ludei

I very much doubt graphical artefacts are related to javascript engines at all. Most likely it's graphics engine responsible for rendering textures.

Are you sure it's not to do with sub-pixel positioning? What if you set sampling to Point and enable pixel rounding in project properties.

Ashley sampling is already set to point and pixel rounding is on. :/

Shameless bump for ludei people.

Why not report this on their own forums?

I thought I couldn't attract attention on their forum because it was pretty much empty but you're right of course.

Today I realized that the amount of inaccuracy in distance depends on the width of the platform object the end sprite's x position is set to. Wider platforms cause greater deviations.