Try Haxe
try.haxe.org
try.haxe.org
Thanks.
Also, the Haxe community is full of great hackers and incredibly helpful.
There are some annoying differences between the different targets but it's >99% the same and you can work around the differences.
Thanks.
Haxe isn't a game engine. It's a language that compiles into other languages.
While compiling to JavaScript obviously results in JavaScript code (which is what the browser can understand), but compiling to Flash does not compile to ActionScript 3 and then the flash assembly (the .swf file). It goes straight from HaXe to the flash assembly.
Likewise, the Java target (which is still in development) won't merely translate HaXe into Java and then compile that Java. I've seen solutions like that and it's not fast or elegant.
Second, I don't see a mention of a physics engine builtin to haXe anywhere on the features page,
where are you seeing that?
Haxe is a language and you can browse the libs available for its compile targets at http://lib.haxe.org/.
If you want to find out more about a framework that uses Haxe and is oriented towards mobile gaming (but not only towards that) checkout HaxeNME.
If you want a nice mature game engine lib built for Haxe try awe6 in the search at http://lib.haxe.org.
Haxe isn't lacking in the game engine department, thing is you can use Haxe with externals for node.js, a lib like noxe that extends haxe to node or simply use whatever javascript lib you feel like and output js. Or use the pure data lib to interact with mobile. Haxe isn't restricted to gaming. It's just one of the things the language allows for.
The Haxe language gives you far more flexibility and performance optimizations, since it is generating optimized native source or byte code. The cost is that you have to learn the Haxe language.
Many people on this discussion are criticizing the java-style syntax of Haxe. I think most of these negative feelings are due to the excessive boilerplate required for many Java classes. Haxe does away with a lot of boilerplate code using type inference and other modern compiler techniques. http://haxe.org/ref/type_infer
You can remove even more boilerplate with macro libraries for shorthand class declarations: https://github.com/back2dos/tinkerbell/wiki/tink_lang
There are many other tricks you can use to define behavior without resorting to complex inheritance or class definitions. One of my favorites is the "using" declaration: http://haxe.org/manual/using
So, in my opinion, the cost of learning Haxe is very low if you already know Java or Actionscript. The cost of coding/writing Haxe is also very low, since you're not required to write boilerplate code.
Finally, it's worth mentioning that the Haxe developers are not just trying to gloss over the technicalities of each underlying platform in order to get things working. They really understand the behavior of each platform, and make many conscientious decisions based on considerations for performance and cross-platform consistency. I've learned more from the Haxe language google group about each target (js quirks, java limitations, etc.) than I have from the Java, JS, Actionscript groups themselves: https://groups.google.com/forum/#!forum/haxelang Many people have remarked that simply following the Haxe development list has made them a better programmer.
http://www.ludamix.com/apps/triad-synth/index.html
Haven't done an official port yet, but one of the NME devs did look into using it to test the SampleDataEvent API on other targets. I did get it compiling but not running, so the ports probably won't be a huge project.
I like a lot of things about Haxe. The syntax is more streamlined than traditional curly-bracers. The type system can massively help with refactoring. It's "honest" about the limitations of targets - you get the APIs pretty much exactly as they are on the target itself. The compiler is well-implemented and updated all the time. It's just a very productive environment, and although it used to have lots of maturity/ecosystem issues, increasingly those aren't a major hurdle, at least for the game space.
Haxe does look great and for this particular case I wonder how Haxe NME's performance compares to the lua frameworks like Corona and Gideros for mobile development specifically.
Flash API to SWF or JS (nme/jeash):
Flash with custom shaders (hxsl):
Flash + away3d:
It may still be a bit more verbose than Coffeescript or Python, but once your code is written you can theorically run it almost anywhere :)
Would like to use this for the game I'm working on since the idea of write once, compile everywhere is seductive. Of course I seem to recall that when Java first came out that was supposed to be it's claim to fame...
It basically provides an implementation of the Flash API for the C++ and Javascript targets, besides the SWF target that Haxe offers natively.
It also comes along with tools that allow this generated C++ code to be easily compiled and deployed to Android, iOS, Blackberry and WebOS devices.
So, if you write an app that targets Flash Player using Haxe, you can use NME to deploy to all these targets with no modification to your code.
I know it sounds superficial, but I'd love a CoffeeScript-like alternative syntax for Haxe. That would probably elevate it to the number one language choice for game programming for me.
The main function has to be static and will serve as the entrance point of the program, not the class. So this might just look like too much when in fact this is just a very small code sample.
Did I address some of your concerns, or did I miss the point entirely? Just curious.
"static function main()" seems to work as an entrance point, though, which makes sense somehow...
enum Cell<T> { nil; cons(car : T, cdr : Cell<T>); }
class List {
public static function mymap<A, B>(f : A -> B, l : Cell<A>) : Cell<B> {
return switch(l) {
case nil: nil;
case cons(x, xs): cons(f(x), mymap(f, xs));
}
}
}
Seriously, if you're prepared to throw out the whole language based on what its syntax resembles, your decision-making process might need a bit of work.
You can compile this example to both Flash and JS online.
If you install Haxe and NME (http://haxenme.org), you can also compile it to iOS, Android, WebOS, BlackBerry, as well as desktop Mac, Windows and Linux binaries.
And these are only the possibilites offered by NME, which is basically a Haxe-based Flash API implementation.
http://www.joshuagranick.com/blog/2012/04/12/html5-pirate-pi...
It's a Bejeweled clone that has been published in the Amazon Appstore, Google Play, BlackBerry App World, the webOS App Catalog and also compiles for iOS, Flash, Windows, Mac, Linux and HTML5
But on the other hand, its syntax and some of its semantics are Java/C++-inspired, whereas something closer to its ML roots would've been so much better.
It might still beat Java for writing Android apps.
The types are based on JS without truthy or falsy values; comparison is required to get boolean values.
Like Coffeescript, while+iterators are used instead of C-style for loops. (no "do while" blocks either)
Case statements don't fall through.
These little changes(plus a few others) help bring in a lot of the enforced-cleanliness that you'd get with a bigger syntax change, it just "looks" curly-brace. One that I miss is :=, since I have run into the ==/= typo occasionally(it just occurs in much more limited situations). No-truthiness is particularly interesting since you only really notice it's there when you try to port C or JS code.
Clément hacked this together as a weekend project. Source code is here for those interested: https://github.com/clemos/try-haxe
Well done :)
In the usual case of using already written cross-platform libs you'd rarely need to do this.
Most standard classes and methods (String, Array, Hash, etc), and a lot of third party libraries are completely cross-platform, so you don't need conditional compilation.
Usually you use conditional compilation either for very low-level per-platform operations / optimizations, or to switch, say, between client code and server code, like here for instance:
http://www.haxenme.org/documentation/examples/tutorials/ http://haxe.org/doc http://www.haxejs.org/ http://www.haxenode.org/ http://mikecann.co.uk/personal-project/more-html5-haxe-speed...
If you do have a go, I recommend HaxeNME for it is a framework that will install all you need for Haxe to just "work" straight up and it also adds some cool stuff like mobile compilation targets (webos, ios, android) and HTML5.
You will have to pick a IDE, people go for FlashDevelop and Mono a lot now, but do read through the Haxe or HaxeNME site setup sections for full list. There are lots of add-ons for a lot of IDE's.
For install, start here: http://www.haxenme.org/documentation/setup/
It was interesting to be able to write code in one language then compile it to other destinations but it seemed to me that targeting specific domains and learning all their ins and outs made more sense.
- Haxe/Flash supports embedding .swf libraries : http://haxe.org/manual/swc
- NME/C++ (Android, iOS, ...) has partial support for .swf libraries : http://www.joshuagranick.com/blog/2012/02/08/quick-notes-on-...
- NME/HTML5 doesn't seem to support .swf right now, but I think it's being worked out.
Then you have third party tools such as Flump or Flambe to help you: https://github.com/threerings/flump https://github.com/aduros/flambe
HaXe is revolutionary, people who really try it always seem to stay with it, they just have to get past the ask lets of questions and actually try some code and stop being negative and putting it in the less able solutions pile. If you want to waste days coding a game in pure objectiveC and then converting it to java just to target two phones then good luck to you! And if you code something in one language it should just be a case of switching some target specific libraries and all your work now works in another target, not recoding in another language style. HaXe is revolutionary because it in achieving what other solutions promise but normally fall short on providing - cross platform cross language, code that is designed to enable not fragment. So try it rather than assume!
You don't have show stopper issues with the compiler ( it outperforms flex compiler on flash target ). Any minor issues are often resolved within hours of reporting, unless they are language design decisions rather than bugs.
Pizzaman
The forum is not well used as most users use the googlegroup 'haxelang'. In terms of frameworks like flex, haxe for flash target does not support flex well, it can be made to work, but largely the problem is because flex is not really just actionscript but has some adobe secrets/hacks etc.. that is why it has not been ported to a cross platform solution yet, need some better engineering decisions although someone I think is working on it. In terms of other frameworks for components there are some but they could be better, try aswing, feel free to chat on IRC if you have problems getting a particular component framework running. Generally haxe is sort of tool that you can build from scratch your visual interface stuff, or if your using js you can mix it up with native libraries.
Jetz
You can add any support for mobile camera's or video you want it interfaces easily with c++/c and there is java support and there is openframework support try this site http://blog.onthewings.net/. Haxe is not really a game only language but it excels at allowing users to create visual stuff but still do complex game mechanics, and a lot of users enjoy doing games. It works well in solving the issues of having to mix technologies, being normally a more sane programming language than the target but also mixing well, so you can mix haxe php, haxe js and haxe flash ( http://haxe.org/doc/remoting ), or use NME library to hit all the phones and have a web version in flash for users to try online.
I think Appcelerator is more adapted to applications that want to use native UIs elements (menus, buttons, layouts, etc).
Then you can still write an Appcelerator app using Haxe :) https://github.com/skial/haXe-titanium-desktop
I don't know how/if it's possible to mix both, for example embed a NME app inside an Appcelerator app (besides using UI.WebView and embed it as HTML...)
Stop fragmenting languages, we don't need another one except if it's revolutionary.
There's some very good principles like the fact that it can compile to many languages but like CoffeeScript I just don't see the point. Want to do some c++ why not using it? Want to do some JavaScript? Then use JavaScript.
There is just too much on the table when it comes to language compilation, I mean if the compiler itself has a problem with your code? It's a bug between what you built and its runtime: what can you do about it? A client is waiting for you, you've got no time to lose contacting the compiler's developers, you're just fucked. Don't get me wrong, I like to play with these languages but that is what they are: toys.
The hard reality is that browsers only support n amount of languages and we are stuck using them.