5 ms·

Aroma - The Native Client game engine powered by Lua

45 points by newtonapple 14y ago

moonboots 14y ago
While I'm concerned that Chrome is the only browser supporting native client (Mozilla outright rejected it), it's still awesome to see lua in the browser.
- oblique63 14y ago
  > Chrome is the only browser supporting native client (Mozilla outright rejected it) Do you have any sources explaining Mozilla's decision on this? I'm curious as to why they would reject it...
  - Jimmie 14y ago
    Publicly they have stated that it's because they think JS is fast enough. I think that they have some private issues regarding a plugin controlled solely by their browser competitor who wants NaCl apps to be purchased through the Chrome Webstore.
    - simonster 14y ago
      NaCl constrains the CPU architectures that the web can run on. No one is going to compile for anything but x86 and ARM. Some people might not even bother to compile for ARM. This seems bad. (If NaCl were widely adopted 5 years ago, the web would be much less accessible to mobile devices.) PNaCl makes things a little better by using LLVM IR instead of machine code, but one of the LLVM developers has publicly stated that "LLVM IR is a poor system for building a Platform" (http://lists.cs.uiuc.edu/pipermail/llvmdev/2011-October/043719.html http://lists.cs.uiuc.edu/pipermail/llvmdev/2011-October/0437...).
      - Jimmie 14y ago
        "If NaCl were widely adopted 5 years ago, the web would be much less accessible to mobile devices." I don't agree with this. From what I've read NaCl isn't even getting auto-run by default. It is not trying to replace JS just sit alongside it as an application language. It lets developers target CPU architectures instead of Operating Systems which makes a lot of sense to me. Also it's C/C++, any popular app would be easily ported to mobile devices. Look how far JS has come in 5 years, imagine how for NaCl could have come. On your PNaCl point, it may not be optimal but I love the idea. "Compile to javascript" is an awful concept, "compile to IR" is better for a number of reasons that I won't go into right now.
        simonster 14y ago
        > I don't agree with this. From what I've read NaCl isn't even getting auto-run by default. It is not trying to replace JS just sit alongside it as an application language. It lets developers target CPU architectures instead of Operating Systems which makes a lot of sense to me. Even if it's not trying to replace JS, if it works well enough, a lot of sites will use it to replace JS, whether because they want to add functionality that's not as easy to implement in JS, because they don't want to expose their source, or just because their programmers don't know/like JS. We have already been through this with ActiveX, which was a similar idea, but poorly executed. > Also it's C/C++, any popular app would be easily ported to mobile devices. Only if the developers care. There's a chicken-and-egg problem here. If your mobile platform can't interact properly with a large proportion of the web, no one is going to buy it, and so no one will bother to port their site to it. > "Compile to javascript" is an awful concept, "compile to IR" is better for a number of reasons that I won't go into right now. Possibly. If you absolutely have to run C/C++ app on the client side, I agree that something like PNaCl (but actually architecture-independent) is a better solution than something like emscripten. But I think it would be even better to 1) run it server side or 2) write it in JavaScript.
  - Arelius 14y ago
    It seems that Brendon has commented on this: http://news.ycombinator.com/item?id=2983157 http://news.ycombinator.com/item?id=2983157 tldr: Nacl cannot be standardized, and the relevant API's are too strongly tied to chrome/webkit details to provide a first-class port to Mozilla.
- afsina 14y ago
  It is Mozilla's failure. Nacl may potentially be the best and widest deployment environment for native code and games ever.
  - zobzu 14y ago
    It doesn't matter, if Firefox supported NaCl none of these games would work in Firefox. They're all Chrome-store only. Besides IE, Opera, etc all rejected it as well.
  - tete 14y ago
    Because what we need is another Active X? The fact that this time it is from Google doesn't make it better. We slowly rid ourselves from stuff like Flash and Java on web pages and now we want to make the same mistakes again? I know, it isn't bad in first place (Flash, Java, Active X weren't either), but I'm pretty sure we'll see better approaches.
    - afsina 14y ago
      I do not recall ActiveX running in MacOsX or Linux. I do not understand when People are completely ok when a game is Windows only but when it is in Chrome suddenly preaches multi-browser compatibility. To me, NACL is not a mistake, it is future. Oh also it is damn open source.
      - hermanhermitage 14y ago
        I'm in two minds, I want a near native code execution environment in the browser but wonder if something like Xax offers better room for performance of dynamic generated code.
    - greggman 14y ago
      The fact that you're comparing NaCl to ActiveX suggests maybe you don't actually understand what NaCl is? NaCl is a platform independent secure code execution environment. It is NOT a way to talk to the underlying OS like ActiveX. Is is NOT a way to write plugins like ActiveX. It's effectively just another way to write code in a browser like JavaScript. Is has no more or less permissions than JavaScript. It is no more or less access to APIs then JavaScript. It just has 5 to 100x more speed than JavaScript depending on what you are trying to achieve.
SomeCallMeTim 14y ago
Moai supports Chrome...and iOS and Android and Windows and Mac: http://getmoai.com/ http://getmoai.com/ It's cool to see Lua in the browser, yes, but it's been there for a while already. I also am skeptical about LÖVE (what Aroma's based on) being ready for commercial games. I've asked LÖVE supporters what commercial games were built on that platform, and they came up completely empty. Bringing a game to release quality tends to find all of the corners of a library that are not production-ready. Moai has a number of serious users, though, including Double Fine.
- kevingadd 14y ago
  I don't think it makes sense to attribute any special amount of quality to Moai; IIRC only one commercial-quality game (Crimson Steam Pirates) has been released in it. DFA will count once it's out, I guess, but that's a ways away. Having used Moai myself I wouldn't really call it commercial quality. The documentation is poor, the code is cryptic, and it tends to crash or do strange things instead of report errors. It also fails to abstract out hardware limitations or notify you of them (for example, it'll let you use 4Kx4K textures and then when you run on a machine with a 2Kx2K limit, Moai just falls over instead of providing clear error feedback or scaling-down the texture.)
  - pygy_ 14y ago
    > only one commercial-quality game (Crimson Steam Pirates) has been released in it. DFA will count once it's out, I guess, but that's a ways away. Add Wolf Toss, and the WIP Shadowrun Returns. Both DFA [Doublefine Adventure] and Shadowrun should be done by the end of the year. Regarding the bugs, I'm sure tat they'd be happy to discuss them, but it's night time in Seattle now. You can drop a line at their forums (http://getmoai.com/forums http://getmoai.com/forums) or the Github bug tracker (https://github.com/moai/moai-dev/issues https://github.com/moai/moai-dev/issues). For the documentation, it is constantly improving. The API docs still have weak points, but the high level overview and the (now partly annotated) examples go a long way. You may want to check these pages: Overview of the runtime: http://getmoai.com/wiki/index.php?title=The_Moai_Runtime http://getmoai.com/wiki/index.php?title=The_Moai_Runtime More details: http://getmoai.com/wiki/index.php?title=Moai_SDK_Basics_Part_One http://getmoai.com/wiki/index.php?title=Moai_SDK_Basics_Part... http://getmoai.com/wiki/index.php?title=Moai_SDK_Basics_Part_Two http://getmoai.com/wiki/index.php?title=Moai_SDK_Basics_Part... Annotations of the examples: http://getmoai.com/wiki/index.php?title=Moai_SDK_Samples_Guide http://getmoai.com/wiki/index.php?title=Moai_SDK_Samples_Gui... TOC: http://getmoai.com/wiki/index.php?title=MoaiSDK http://getmoai.com/wiki/index.php?title=MoaiSDK
tferris 14y ago
Why is Lua so popular among game devs?
- Jimmie 14y ago
  It's fast, dynamic and easy to embed in a C/C++ game for scripting. The language itself is very simple to learn too.
- kevingadd 14y ago
  It has most of the positive traits of JavaScript with less of the negative traits. A few examples: Strings in Lua are interned, which makes it possible to efficiently use string keys in all sorts of places instead of numeric enumerations. Lua's garbage collector has support for running incrementally with a configurable pause time, so you can run it in the background every frame to avoid those nasty GC pauses you still get in most modern browser games from stop-the-world collections. Debugging is built into the language with a user-accessible API that lets you write your own debugger or diagnostic tool, in addition to using third-party ones, instead of having to rely on the vendor to provide one. Lua uses a single container type for storing data, with an efficient representation (densely packed elements are stored in an actual array; sparsely packed elements and elements with non-integral keys are stored in a hashmap). As a result, operations on sequences/containers only ever have to think about one data type, and you still get the near-optimal performance you would in JS if you manually used Array/Object where most appropriate.
  - lucian1900 14y ago
    Actually, JS Arrays are implemented as hashmaps as well.
  - simonster 14y ago
    > Strings in Lua are interned, which makes it possible to efficiently use string keys in all sorts of places instead of numeric enumerations. Interning is an implementation detail. I don't know how they are implemented off the top of my head, but string keys are pretty damn fast in most JS JITs, since they're used to get e.g. methods of objects. > Lua's garbage collector has support for running incrementally with a configurable pause time, so you can run it in the background every frame to avoid those nasty GC pauses you still get in most modern browser games from stop-the-world collections. Again, this is an implementation detail. Firefox is about to get incremental GC. While the pause time isn't configurable, it doesn't really matter as long as it's short by default. > Debugging is built into the language with a user-accessible API that lets you write your own debugger or diagnostic tool, in addition to using third-party ones, instead of having to rely on the vendor to provide one. Gecko and WebKit both have debugging APIs. I will concede that it's a PITA that they're different. > Lua uses a single container type for storing data, with an efficient representation (densely packed elements are stored in an actual array; sparsely packed elements and elements with non-integral keys are stored in a hashmap). As a result, operations on sequences/containers only ever have to think about one data type, and you still get the near-optimal performance you would in JS if you manually used Array/Object where most appropriate. Again, this is an implementation detail. In JS, Array is a subclass of Object and in most JITs I think it's implemented the same way. The only difference is how they serialize to JSON and that Array has a length property. Lua has an advantage in that it gives arrays both sparse and dense slots, but this isn't necessarily impossible to do in JS either (https://bugzilla.mozilla.org/show_bug.cgi?id=586842 https://bugzilla.mozilla.org/show_bug.cgi?id=586842)

45 points by newtonapple 14y ago

While I'm concerned that Chrome is the only browser supporting native client (Mozilla outright rejected it), it's still awesome to see lua in the browser.

oblique63 14y ago
> Chrome is the only browser supporting native client (Mozilla outright rejected it) Do you have any sources explaining Mozilla's decision on this? I'm curious as to why they would reject it...
- Jimmie 14y ago
  Publicly they have stated that it's because they think JS is fast enough. I think that they have some private issues regarding a plugin controlled solely by their browser competitor who wants NaCl apps to be purchased through the Chrome Webstore.
  - simonster 14y ago
    NaCl constrains the CPU architectures that the web can run on. No one is going to compile for anything but x86 and ARM. Some people might not even bother to compile for ARM. This seems bad. (If NaCl were widely adopted 5 years ago, the web would be much less accessible to mobile devices.) PNaCl makes things a little better by using LLVM IR instead of machine code, but one of the LLVM developers has publicly stated that "LLVM IR is a poor system for building a Platform" (http://lists.cs.uiuc.edu/pipermail/llvmdev/2011-October/043719.html http://lists.cs.uiuc.edu/pipermail/llvmdev/2011-October/0437...).
    - Jimmie 14y ago
      "If NaCl were widely adopted 5 years ago, the web would be much less accessible to mobile devices." I don't agree with this. From what I've read NaCl isn't even getting auto-run by default. It is not trying to replace JS just sit alongside it as an application language. It lets developers target CPU architectures instead of Operating Systems which makes a lot of sense to me. Also it's C/C++, any popular app would be easily ported to mobile devices. Look how far JS has come in 5 years, imagine how for NaCl could have come. On your PNaCl point, it may not be optimal but I love the idea. "Compile to javascript" is an awful concept, "compile to IR" is better for a number of reasons that I won't go into right now.
      - simonster 14y ago
        > I don't agree with this. From what I've read NaCl isn't even getting auto-run by default. It is not trying to replace JS just sit alongside it as an application language. It lets developers target CPU architectures instead of Operating Systems which makes a lot of sense to me. Even if it's not trying to replace JS, if it works well enough, a lot of sites will use it to replace JS, whether because they want to add functionality that's not as easy to implement in JS, because they don't want to expose their source, or just because their programmers don't know/like JS. We have already been through this with ActiveX, which was a similar idea, but poorly executed. > Also it's C/C++, any popular app would be easily ported to mobile devices. Only if the developers care. There's a chicken-and-egg problem here. If your mobile platform can't interact properly with a large proportion of the web, no one is going to buy it, and so no one will bother to port their site to it. > "Compile to javascript" is an awful concept, "compile to IR" is better for a number of reasons that I won't go into right now. Possibly. If you absolutely have to run C/C++ app on the client side, I agree that something like PNaCl (but actually architecture-independent) is a better solution than something like emscripten. But I think it would be even better to 1) run it server side or 2) write it in JavaScript.
- Arelius 14y ago
  It seems that Brendon has commented on this: http://news.ycombinator.com/item?id=2983157 http://news.ycombinator.com/item?id=2983157 tldr: Nacl cannot be standardized, and the relevant API's are too strongly tied to chrome/webkit details to provide a first-class port to Mozilla.
afsina 14y ago
It is Mozilla's failure. Nacl may potentially be the best and widest deployment environment for native code and games ever.
- zobzu 14y ago
  It doesn't matter, if Firefox supported NaCl none of these games would work in Firefox. They're all Chrome-store only. Besides IE, Opera, etc all rejected it as well.
- tete 14y ago
  Because what we need is another Active X? The fact that this time it is from Google doesn't make it better. We slowly rid ourselves from stuff like Flash and Java on web pages and now we want to make the same mistakes again? I know, it isn't bad in first place (Flash, Java, Active X weren't either), but I'm pretty sure we'll see better approaches.
  - afsina 14y ago
    I do not recall ActiveX running in MacOsX or Linux. I do not understand when People are completely ok when a game is Windows only but when it is in Chrome suddenly preaches multi-browser compatibility. To me, NACL is not a mistake, it is future. Oh also it is damn open source.
    - hermanhermitage 14y ago
      I'm in two minds, I want a near native code execution environment in the browser but wonder if something like Xax offers better room for performance of dynamic generated code.
  - greggman 14y ago
    The fact that you're comparing NaCl to ActiveX suggests maybe you don't actually understand what NaCl is? NaCl is a platform independent secure code execution environment. It is NOT a way to talk to the underlying OS like ActiveX. Is is NOT a way to write plugins like ActiveX. It's effectively just another way to write code in a browser like JavaScript. Is has no more or less permissions than JavaScript. It is no more or less access to APIs then JavaScript. It just has 5 to 100x more speed than JavaScript depending on what you are trying to achieve.

SomeCallMeTim 14y ago

Moai supports Chrome...and iOS and Android and Windows and Mac: http://getmoai.com/ http://getmoai.com/ It's cool to see Lua in the browser, yes, but it's been there for a while already. I also am skeptical about LÖVE (what Aroma's based on) being ready for commercial games. I've asked LÖVE supporters what commercial games were built on that platform, and they came up completely empty. Bringing a game to release quality tends to find all of the corners of a library that are not production-ready. Moai has a number of serious users, though, including Double Fine.

kevingadd 14y ago
I don't think it makes sense to attribute any special amount of quality to Moai; IIRC only one commercial-quality game (Crimson Steam Pirates) has been released in it. DFA will count once it's out, I guess, but that's a ways away. Having used Moai myself I wouldn't really call it commercial quality. The documentation is poor, the code is cryptic, and it tends to crash or do strange things instead of report errors. It also fails to abstract out hardware limitations or notify you of them (for example, it'll let you use 4Kx4K textures and then when you run on a machine with a 2Kx2K limit, Moai just falls over instead of providing clear error feedback or scaling-down the texture.)
- pygy_ 14y ago
  > only one commercial-quality game (Crimson Steam Pirates) has been released in it. DFA will count once it's out, I guess, but that's a ways away. Add Wolf Toss, and the WIP Shadowrun Returns. Both DFA [Doublefine Adventure] and Shadowrun should be done by the end of the year. Regarding the bugs, I'm sure tat they'd be happy to discuss them, but it's night time in Seattle now. You can drop a line at their forums (http://getmoai.com/forums http://getmoai.com/forums) or the Github bug tracker (https://github.com/moai/moai-dev/issues https://github.com/moai/moai-dev/issues). For the documentation, it is constantly improving. The API docs still have weak points, but the high level overview and the (now partly annotated) examples go a long way. You may want to check these pages: Overview of the runtime: http://getmoai.com/wiki/index.php?title=The_Moai_Runtime http://getmoai.com/wiki/index.php?title=The_Moai_Runtime More details: http://getmoai.com/wiki/index.php?title=Moai_SDK_Basics_Part_One http://getmoai.com/wiki/index.php?title=Moai_SDK_Basics_Part... http://getmoai.com/wiki/index.php?title=Moai_SDK_Basics_Part_Two http://getmoai.com/wiki/index.php?title=Moai_SDK_Basics_Part... Annotations of the examples: http://getmoai.com/wiki/index.php?title=Moai_SDK_Samples_Guide http://getmoai.com/wiki/index.php?title=Moai_SDK_Samples_Gui... TOC: http://getmoai.com/wiki/index.php?title=MoaiSDK http://getmoai.com/wiki/index.php?title=MoaiSDK

tferris 14y ago

Why is Lua so popular among game devs?

Jimmie 14y ago
It's fast, dynamic and easy to embed in a C/C++ game for scripting. The language itself is very simple to learn too.
kevingadd 14y ago
It has most of the positive traits of JavaScript with less of the negative traits. A few examples: Strings in Lua are interned, which makes it possible to efficiently use string keys in all sorts of places instead of numeric enumerations. Lua's garbage collector has support for running incrementally with a configurable pause time, so you can run it in the background every frame to avoid those nasty GC pauses you still get in most modern browser games from stop-the-world collections. Debugging is built into the language with a user-accessible API that lets you write your own debugger or diagnostic tool, in addition to using third-party ones, instead of having to rely on the vendor to provide one. Lua uses a single container type for storing data, with an efficient representation (densely packed elements are stored in an actual array; sparsely packed elements and elements with non-integral keys are stored in a hashmap). As a result, operations on sequences/containers only ever have to think about one data type, and you still get the near-optimal performance you would in JS if you manually used Array/Object where most appropriate.
- lucian1900 14y ago
  Actually, JS Arrays are implemented as hashmaps as well.
- simonster 14y ago
  > Strings in Lua are interned, which makes it possible to efficiently use string keys in all sorts of places instead of numeric enumerations. Interning is an implementation detail. I don't know how they are implemented off the top of my head, but string keys are pretty damn fast in most JS JITs, since they're used to get e.g. methods of objects. > Lua's garbage collector has support for running incrementally with a configurable pause time, so you can run it in the background every frame to avoid those nasty GC pauses you still get in most modern browser games from stop-the-world collections. Again, this is an implementation detail. Firefox is about to get incremental GC. While the pause time isn't configurable, it doesn't really matter as long as it's short by default. > Debugging is built into the language with a user-accessible API that lets you write your own debugger or diagnostic tool, in addition to using third-party ones, instead of having to rely on the vendor to provide one. Gecko and WebKit both have debugging APIs. I will concede that it's a PITA that they're different. > Lua uses a single container type for storing data, with an efficient representation (densely packed elements are stored in an actual array; sparsely packed elements and elements with non-integral keys are stored in a hashmap). As a result, operations on sequences/containers only ever have to think about one data type, and you still get the near-optimal performance you would in JS if you manually used Array/Object where most appropriate. Again, this is an implementation detail. In JS, Array is a subclass of Object and in most JITs I think it's implemented the same way. The only difference is how they serialize to JSON and that Array has a length property. Lua has an advantage in that it gives arrays both sparse and dense slots, but this isn't necessarily impossible to do in JS either (https://bugzilla.mozilla.org/show_bug.cgi?id=586842 https://bugzilla.mozilla.org/show_bug.cgi?id=586842)