hmm. This is an extremly important question for me as well.. well will be.

The CPU cycles are less important that the effect of deterministic physics. If C2 uses deterministic physics then the CPU should make no difference. Unfortunatly Box2d is not natively built as deterministic.

The current solution is to run the physics step independently of the main gameloop and instead run the step based on timestamps. However, it seems to me that in C2 there isn't a way to turn on or off the physics(I could be wrong I haven't delved into c2 physics that much).

In the mean time you might have to do Whiteclaws suggestion. Get the difference in FPS. Determine a base value for impulse. Then figure the percentage of change based on the FPS difference. I suspect the physics simulation won't be the same. There is consideration for friction, density and mass for the fixture where impulse has different effects :|

Thanks. I really appreciate you showing the drag and drop behaviour. I think it works out really well :) I think the tutorial will be a lot better for your contribution once it's updated :)

You rock :)

Could you please put up an HTML sample. I don't use beta versions and I want to test this out on an actual touch screen.

Honestly I completely overlooked the Drag and Drop behaviour. Though I tried to simulate it by doing the same form of checking to see if the touch was still touching. I found that it was unusable as the touch would lose it's requirement of still touching thumb stick. However maybe the drag and drop behaviour will be more solid in monitoring the touch position even when the finger slides off the thumb stick.

just a hunch.

Have your oblong object. Place the center of mass at the bottom. Make sure the collision has a rounded bottom. THen make sure it's heavy enough.

hmm. there might be a problem with your explanation there might not be. But I'll try to give some experience and some advice anyways.

If your trying to get true metric/imperial sizes for screen size. Don't even try. I tried this myself. Supposedly browser/screen makers should be giving out accurate dpi for programs to offer a way to get "true" sizes. They don't. You can create a program or anything that has say 5 centimeters in measurement. You will never get that 5 centimeters. I went around 4 different size monitors, 4 different sized mobile devices(10", 9", 7", 3.5") and my 1 inch comparison never really matched up to the ruler. Larger monitors were far closer though. This is not a issue with C2 as it is with browser makers or possible screen manufacturers.

scirra.com/tutorials/37/beginners-guide-to-construct-2/page-7

OK. so I gave this one some more thought. rather than store every tick of directional input. Store each change of input with a timestamp. That way you have to store less and can check to see if the time stamp is within the range of the move.

Your going to need to store your input into a array or some list that you can look back on. Create an array that can store about 1/2 seconds of input.

Every input should be stored into the array.

When a movement key input(like punch) is pressed. Then track the the movement backwards to make sure only wanted input is used.

Your best off I think with your second solution. If your sprite is THAT big. chop it up and place it in a grid to be displayed. Let C2 do the culling work for object outside the viewport. Also when chopping your map make sure to chop them into the standard binary exponent.

1x1

2x2

4x4

8x8

16x16

32x32

64x64

128x128

256x256

512x512

1024x1024

Also don't forget that 0 counts as 1. so it would be 255x255

ummm. well. A thought would be to do a "check on every X seconds" say 1. If your fortunate the window Width will change. Check all the Width of course just to make sure. If it changes then you have your check. I used this model in straight js, but i'm guessing here in C2.

oh oh. You should be able to check the acceleramotr instead of width. I'm not sure which vector your checking for or the degree's before the tablet instelf changes. but I think this will work out.

Thanks. Not as elegant as knowing how to do the angle system and math checks, but it was WAY easier to implement and works. I popped two empty sprites at the center location. pinned and rotated them to 90 degress(relative) and then used the same event as I did with the working move forward. Worked great on the first go.

So i'm playing with my controls and trying a "swipe" motion based movement system.

I'm having troubles with the angle system since the motion system uses relative angle checking. C2 uses -180 to 180 making simple angle checking difficult.

Image here(   https://docs.google.com/drawings/d/1auRRMO3CokJ-wqP7X104lPED74fcx03W6zs-pK7Dlsc/edit   )

Image

There are 3 samples of an arrow with the relative direction.

The control works by having the player put their finger on the arrow first. Then release the finger in the relative direction.

Forward works well as I can use "Within". However, trying to use the relative angle for left and right turning is not working at all. well it works when the object is in the original position as that is the angle.

So is there a way to use within to check an angle that isn't the angle of the arrow or is there another way to handle how the angle system works.

I haven't done this myself, but I have had the same thought about it. At some point I wanted to do some kind of simple old school 2d Zombie GTA kind of thing. I figure the project would be fairly large so I never started. But I did ponder about the large map issue as I felt that would be the most difficult hurdle.

As I approached the theory I cam up with two ideas. I don't claim these as only or best solutions. Just ideas based on how much I understand C2.

One. C2 functionality

Have a two large set of empty sprites of a variety of sizes and ID/Use types.

Buildings, Cars, Roads, Sidewalks.... These empty sprites are always the same name. Your dynamic sprites are managed as two types(A and B) to manage sprite loading for when changing the environment.

Define a system to alternating URL load different sprite set's based on the players position in city grid.

Two. Plugin

That does the above, but removes a very complicated Event Sheet.

Anyways. that's my two thoughts on the subject. I would like to see someone pull off the really big map idea. I may end up tinkering on my project sooner than later :P