Posts by cgbartek

I can't answer for SirEatAlot, but I've been sticking with XCode 4 since I use Synergy quite a bit and it does not have Mavericks support yet (and XCode 5 requires Mavericks).

However, I think Apple requires all new projects be compiled with XCode 5 now, so I guess I'll just have to bite the bullet and find another KVM solution for my next project.

I actually reported this exact problem to the bug tracker some time ago.

There are two ways to solve this:
1) Make sure the complete path to your project has no spaces in it, or
2) In Xcode, go into Targets > Search Paths and add quotes around the Header Search Paths and User Header Search Paths. Do this for both your project and the Box2D project.

Either of those will fix your problem.

From what I understand the Runtime object was phased out awhile ago, and using it will cause problems in some runtimes (iOS for instance). It's a shame, since it was an extremely useful extension.

I can confirm the bugs with layers appearing and disappearing in iOS, so it's definitely not just you. I also posted my similar problems to the bugbox.

Awesome, glad that fixed it. You're correct, CORS should not be an issue when you run the game from your own server, but for development reasons is will be a necessity. Modern web browsers consider it a security risk to send requests to external servers that don't have CORS enabled; MMF executables, etc. don't respect those concerns, so they just work normally.

Are the scripts on your server set up to use CORS? That fixed it for me.
Please login to see this link.

Is the server returning blank responses, or MMF?

I was about to send the mfa over, but it seems to be working now. Perhaps closing and re-opening the project got it working? I'll keep my eye out of this problem happens again.

Some bugs, but first, I just wanted to say that at this point the runtime is remarkably bug-free compared to the other runtimes I've been using. This is all around spectacular work so far, and I honestly had my doubts that a fast, reliable HTML5 runtime was even possible. Thanks Francois for the hard work you've put into this runtime.

Beta 16 bugs:

- If you have an unsupported object in the project (Android Object, etc.) the final build will not work. Error: "An error has occurred while building the Javascript code. The HTML5 project could not be built". It would be good to be able to compile with unsupported objects for cross platform project reasons.

- I can't seem to get an object's alpha coefficient. Compatibility semi-transparency works fine.

- Previously reported bug: if an object is hidden at start on a hidden layer, it cannot be un-hidden. Only workaround is to have the layer visible at start, then immediately hide it. Bug also exists in the Flash runtime.

- One of my games refused to run in Chrome, until I moved the frames to a new project. The only thing I can think of that could have been causing the problem, is maybe it thought WebGL was still on, even though the option is unavailable now.

That's all I can find so far. If these get fixed all my projects will all be running at 100%! I'm actually running out of source material to find bugs in.

Make sure you've enabled CORS in your PHP script, otherwise your server might reject a cross-domain request. Just add this to the first line in your PHP file:
header("Access-Control-Allow-Origin: *");

Also make sure you've added the INTERNET permission in your application's Android settings.

This is impressive, but the research document states that it does not run in real-time. And maybe it's just me, but it doesn't look like much of an improvement over hq4x.

Unfortunately I don't have a snippet for decryption, but I would assume it should work roughly the same, except running mcrypt_decrypt on an encrypted string taken from the $_GET or $_POST. The padding thing may or may not be necessary, and you probably need to unhex the string before decrypting. This should point you in the right direction: Please login to see this link.

Quote from MattEsch

Ok I'm going to have to dispell the idea that HTML5 isn't ready and it is slow and other such nonsense.

First of all, I will admit that quality does vary wildly, depending on choice of browser. I'm not suggesting that you give the user the choice of browser, but you a) write extensions using JavaScript b) provide a lightweight container so that it runs fast as a native application in the target environment.

In fact, I would write an object-based rendering engine, capable of rendering textured surfaces, SVG and 3D scenes (I have been working on this anyway), and use JavaScript in V8 to drive it.

I am saying, kill runtime bytecode interpretation. Downcompile to JavaScript, run it on the fastest JS implementation there is (which compiles to assembly code on the fly) and use it to drive a graphics engine implementation, which could be WebGL based or based on DirectX or OpenGL, but ideally what is a single executable and not something which requires a browser to run.

To make games work with JavaScript you need to control the environment in which the user runs it, period.

HTML5 is suitable for mobile scenarios. It's awful in the default browsers, but you wrap your game in a native, hardware accelerated container (See Please login to see this link.).

I only suggest JavaScript because it's ubiquitous, it has momentum and it's only going to improve. It's embeddable and demonstrably capable of driving a modern game, just as Lua or Python (Anaconda is still the most efficient runtime for MMF2), and other scripting languages can. The expense is in the graphics rendering, so providing you have a native hardware accelerated context in which to render things, it works.

I also argue HTML5 is the simplest way to build user interfaces today, and as a product that aims to be intuitive , you need to give your developers the best chance of building exactly what they want, looking exactly how they designed it, as fast as possible.

Display More

Some good thoughts here. HTML5 will eventually become a great way to distribute cross-platform games, much like Flash. However, I think the biggest problem with ditching native runtimes and staying only with HTML5, is HTML5 has numerous low-level limitations (such as file and system access) that would make things impossible for application developers (not everyone makes games with MMF ). I guess APIs could be written into the wrapper ala Metro apps, but that might make things complicated for extension developers. All that said, I think HTML5 is the most important runtime that Clickteam will ever create, and it should remain high on the priority list of runtimes to maintain.

On a side note, have you managed to get CocoonJS to work with MMF? PM me if you have, thanks

No problem tekila, glad I could help.

- 1) Create a second layer, 2) uncheck Visible at start on layer, 3) place an active on that layer, 4) also uncheck Visible at start on object. Go to event list, 5) unhide layer 2 with layer object, 6) now try to unhide the active on the next line, then run... it doesn't work. NOTE: This bug also exists in the Flash runtime.

This is how I got PHP to pass encrypted data to an application using the GET object:

$password = "mustbe16charactr"; //MUST be 16 characters
$text = "This is a test.";


function aes_encrypt($val,$ky) {
    $mode=MCRYPT_MODE_ECB;    
    $enc=MCRYPT_RIJNDAEL_128;
    $val=str_pad($val, (16*(floor(strlen($val) / 16)+(strlen($val) % 16==0?2:1))), chr(16-(strlen($val) % 16)));
    return mcrypt_encrypt($enc, $ky, $val, $mode, mcrypt_create_iv( mcrypt_get_iv_size($enc, $mode), MCRYPT_DEV_URANDOM));
}


echo strtoupper(bin2hex(aes_encrypt($text,$password)));

Basically what has to happen, is the returned encryption must be padded out, then turned to uppercase hex. Then it'll work like a charm in AES Fusion.

I love this extension! This can definitely be the de-facto data storage object for MMF, and then I won't have to use AssArray anymore. Some suggestions:

1) Remove All option to clear the whole thing
2) XML/JSON/INI Import/Export (including a way to get/set the file data as a string so it can be used in conjunction with the GET object)
3) AES encryption/decryption of import/export
4) Port to other platforms

The big thing here would be JSON import/export as a string. This would be a massively, massively awesome thing to have for those of us building web connected applications.

- It is common to get the "Please build your application as a HTML5 project before using the Build & Run option." error when opening a second project and compiling it. Restarting MMF fixes the problem.
- Running RGBAt on an active object crashes the frame.
- Replace color on an active object does not appear to work (I don't know if this feature might be too difficult to make work in HTML5, if so, not a big deal)
- Load frame not working (again, not sure if it's possible in HTML5 but would be nice)

More bugs:
- If you have a fade-in transition on your frame, and a "Start of Frame" event, the frame will crash. Workaround: if you either delete the actions from the event or change the event to "Timer Equals 00"-01", the frame will work again.
- If you hide or show the high score table, the frame freezes/crashes.
- Create a Date/Time object, make it a Clock, set it to Digital, set Face to "01:27:18" setting. Frame will crash at start. The other settings seem to work fine.
- Alpha coefficients don't seem to default to 0. Trying to add to the co-efficient does not fade out the image as expected. Workaround: set the co-efficient to 0 manually at frame start.
- Create a counter, set it to "Text" mode. The display seems to be cut off and only displays the first digit.

Quick note on the GET object: I was able to make it work on external domains by enabling CORS on my backend script. Here's a great website that tells you how to do this: Please login to see this link.

Beta 7 problem:
Okay, this is a weird one. I have an application where any frame with a backdrop or quick backdrop in it will crash the HTML5 app. If you create a new frame, then move everything into it, it will magically fix itself. So it's something to do with the frame, but it doesn't look like there are any frame settings different from the old one. I have attached an example project with 2 identical frames... one works, and the other does not. Drag the non-working frame above the working one so it boots first, build, and see what I mean. Weird eh?

Please login to see this attachment.

+1, this extension is really useful. Would love to see this in the HTML5 runtime as well.

What I've had to do is create an MD5 out of the data I'm sending, plus a salt, and pass that along with the GET string. Then the server script does the same thing and checks to make sure the data hasn't been modified. AES would make it more secure, though, for sure.