GameClosure HTML5 SDK just went open source

I'm not sure what's to gain with this over something like CocoonJS. Unless they support WebGL (standards-compliant, and not a custom OpenGL binding), I don't see how it would be faster, or easy to integrate.

Performance aside and I will admit it might not be faster. The developers say it's an OpenGL graphics context and not art of the standard 2d one. I have no idea what what CocoonJS runs on, but no denying CJS last 1.3 performance gain is really good. So performance aside I don't know which is faster.

however I think the biggest opportunity isn't the performance gain in comparison to CocoonJS, but instead it's open ability for community extensions. Ludie mentioned an extension manager, but no ETA. This could be months or years. Though I have to admit a plus is that CoJS offers already a lot of features.

On a final note. If CocoonJS is the defacto use. Why bother with PhoneGap or AppMobi? Which are still in the exports.

Hi,

I was actually offered an all-expenses paid job at Game Closer company.

(did not want to move from New York to California so I declined)

They are still hiring if anyone is interested.

JeZ+Lee

Hi,

I was actually offered an all-expenses paid job at Game Closer company.

(did not want to move from New York to California so I declined)

They are still hiring if anyone is interested.

JeZ+Lee

They only seek developers who live in California, not freelancers.

So I spent the day trying to install GameClosure.

My skills at Posix/Unix/Bash just aren't up to snuff to get a OSX focused set up to work on Windows. However there is another person that is managing it. If they do it I will take a look and try to see if a minimal canvas pass will work.... or someone can beat me to it whenever they want. I'm not exactly quick, infact I'm pretty slow :D

Barista however is another concern. Barista is a kind of object language that is the model for community extensions. So to create an Extension is to use Barista. This actually adds a lot more overhead to just get access to native tools.

GameClosure might be good to look at once it has matured a little more and someone has managed to get an easy to Windows functionality. In the mean time I will just keep an eye on it :) unless of course you are a Mac developer :)

Decided to give it another go. Spent a number of hours on working to see if C2 could easily work in GC. Of course this was at the expense of my own game :D

So I managed to figure an easy way to get a C2 game running in GC. Requires a few simple hacks, but there is a lot I don't understand to make it less hacky. Also promisingly ran to run in WebGL mode on the initial test, but failed. This was promising, but I didn't want to tinker with OGL yet. In Canvas2D it ran.

As for mobile peformance. I have no idea. Much of GC is BASH command line. Requiring knowledge I'm not familiar with. Requiring a link to the Android SDK the through BASH which is there. So currently all Android GC kit isn't installed. But if anything comes up in the GC community answering how to. I will post an update on performance asap.

As I look some of the examples in this video, all I can say is wow. I think that "photon benchmark" test shows some true power, with over 8000 sprites. I think this would be perfect fit for Construct 2 because it's open source and who knows could be used as an exporter bundled with C2.

http://doc.gameclosure.com/

well as an update. I managed to get an Android compile. There was a lot to go over to get it working on Windows. until they get there Windows scripts going I don't think I could suggest it. The number of hacks are a lot. But I managed it. This includes working with GC and the Android.

Unfortunatly the apk program didn't quite work. it compiled and ran, but stuck on the GameClosure splash screen. I figured I would let it drop for now and focus on my game. Then once that was done come back to GC and figure it out :D

I saw the 8000 sprites. It will be impossible for C2 to get those results. C2 is designed around Canvas Rendering. Where as those 8000 sprites are designed around efficient OpenGL rendering. To take advantage of that Rendering performance I suspect that it would require using the the natural GameClosure Game api in OpenGL layer. That would require an entire new language exporter.

Even if we were to get a C2 WebGL/Layer to get access to the OpenGL canvas that GC provides. I suspect because C2 doesn't work on efficient GL texture and draw call bundling that gives OpenGL it's performance edge. I could be wrong on C2 I don't know how it works. it's just an impression as I'm pretty sure Canvas2D performance design is not the same as OpenGL performance design. again I could be wrong. However it doesn't mean we won't see a large performance increase if it were to work :)

and the benefit of performance AND extendibility is a win in my books. but let's just see where it goes :)

I don't think we could hit 8000 sprites anyway, as I think c2's sprites have a lot more overhead than theirs do in that vid (c2's sprites have variables, behaviors, etc). They might even be using some sort of quick blitting technique I've heard people talk about, keeping the image locations in an array or algorithm or some such then drawing an instance at each location instead of having a fully fledged object at each location, which would result in no extra overhead at all per instance.

I did a test to find out how many invisible sprites with bullet and wrap behaviors I could get in a layout at 55fps:

Ipad 3, safari: 1711, cjs: 770

Iphone 4s, safari: 2197, cjs: 642

So it seems even with rendering entirely off it wouldn't get anywhere near that high. Although I suppose they could have that test running on some crazy cutting edge mobile hardware, but I don't think something like an iPhone 5 could manage it.

hmm. iPad 3 huh. Well I know in the Ouya tests someone managed to get 5000 sprites. That's on a Tegra 3. youtube.com/watch

You are right. There is more overhead. I suspect the test is just a simple brute force object moving. There is no game code overhead. I noticed your cjs tests are slower. Maybe because the js is not running on a more efficient JIT. I hear nitro does a good job to improve script performance.

GameClosure developer has released a port of C2 games to GameClosure.

Check this out github.com/bubbleboy14/scblaster

It seems he is working to make it stable.

Wow, that's pretty awesome.

yeah. i'm glad some one at GC found a solution. Knowing the in's and out is easier to create the bridge.

I'm looking forward to seeing how GC and PhoneGap will be for Ouya come end of March and I can start testing on my own device :)

He has been working on this a lot over the last few days. That's a good sign.

github.com/bubbleboy14/scblaster

Ashley: Gameclosure is open-source and free, while Cocoon isn't (or won't be). Cocoon's pricing model might not suit everyone. It would be great if Construct 2 would support Gameclosure export as well!

It's also helpful to note that GC is now supporting windows properly. When this thread started it was only OSX :D

I personally think Ashley may support it if there enough basic operation on GC that Ash doesn't need to code much :P

Currently as I understand the situation Mario of GC developed a C2 front of the application.js(?) that covers everything needed to run a C2 games except audio and file structure.

Right now the audio manager needs to be handed coded into the appx.js. Which is an array structure of the audio files used in the C2 project.

Also a proper C2 export would need to organize the resources files into a different directory structure.

gameclosure.com/blog/2013/03/space-blaster-meets-mobile

personally I agree. Having an accelerated canvas wrapper that isn't licensed is awesome. But I do admit Ludie CJS is farther down the road on features and coming out soon'ish with WebGl.