Jump to content

hawk66de

Member
  • Content Count

    48
  • Joined

  • Last visited

  • Medals

Community Reputation

10 Good

About hawk66de

  • Rank
    Lance Corporal
  1. Hey, would like to know if some guy here has built or has ideas about a unit testing framework for Arma script? I mean a very lightweight one...not like NUnit for example...
  2. hawk66de

    AI not in foucs ?

    you're right^^we should come back to the OT. So, I'd really like to know if there is a 'common' agreement that the AI guys should get more individual personality...so bringing a little rpg to make them less robotic? Now I remember the project title: "Next Generation PC Game". And some of the features, which I have describe in my initial post (role playing stuff, individual AI personality) were on the feature list.
  3. hawk66de

    AI not in foucs ?

    Hasn't the PS3 -for example- 8 cores and six are accessible concerning game development? I think a major problem is that parallel programming is still in its infancy...but to be fair Arma 2 uses already one core exluslively for AI...think that's pretty (still?) unique. My point was not so the pathfinding issues but more the higher level AI, also including the rpg aspects
  4. hawk66de

    AI not in foucs ?

    Ok, then I might got it wrong...does anybody have a link with an interview of him where he talks a bit about planned AI stuff and so on?
  5. hawk66de

    AI not in foucs ?

    I've read an interview with Ivan Buchta http://www.gamestar.de/spiele/arma-3/artikel/arma_3,46950,2560283.html and one of his statement is that in the planned campaign the player would almost be able to win 'alone', that is without any AI support (if he wishes to do so). That sounds to me, unfortunately, that AI seems not to be the focus of the new release and that the new campaign design in this respect is somehow a 'workaround', reacting on the critics of Arma II regarding this aspect. The question is if this is a desirable solution? Before Arma II was announced, weren't there some ideas from BIS that they create a real unique AI also in terms of personality and introducing role-play aspects (dynamic chatting with AI soldiers, also about personal stuff, which influences moral and so on)? I have really liked those ideas some years ago because they would bring this (great) series to a new level. I do not know what the community thinks but for me AI is the most important aspect of simulating a digital battlefield. But, perhaps it's related to the market situation, where you still sell a game due to it's graphics/content etc and not due to its AI. A couple of months ago I've listened to a podcast about game development. The guy (an EA game developer/architect) claimed that in future the game industry would shift its focus from visuals to behaviors (AI related stuff) because concerning visuals there wouldn't been substantial improvements possible and it would be hard to create an unique product, which surpasses the competition if not doing so. I still doubt that we have reached this state...also if you see that more and more folks play 'casual' games on their iPads and smartphones with minimal AIs.
  6. Concerning the usage of SQF/SQS: It is surely not used only for mission making, but also to improve the engine in general, or to make total conversion mods. I also did not claim that it is not possible to build libraries with SQS...my point is that it is just c like header files but not strong libraries in sense of OOP/OOD I understand that many folks wanna keep the sqf/sqs because they are familiar with it...but hey think about that: assume every product/game, which is extensible/moddable would use its own property language and concepts...do you really wanna learn them 'all' and use your spare time for that? How do you wanna create more and more complex worlds/missions when not the tools and languages support you in doing so and get also considerably upgraded? Don't you think that the complexity of the language/concepts (think about the different concepts like FSM, SQF/SQS, function modules, different config files, hundreds, distinct commands etc.) will reach a state where only a handful, hard core modders can and want handle that? I mean not all the people, who want to to contribute or make sth with it, have simply the time to collect all necessary infos here in the forum, wiki etc. and try it out the hard way. Domain specific languages like Unreal Script are also a valid approach, but then the prerequisite is that you have the same or even a higher productivity for the domain you using them. Again, the arma engine is very capable. My claim is just that using a complete redesigned domain specific language or switching to an industry standard full blown language will make this engine even much more awesome in the longterm.
  7. I fully agree with the topic owner. I don't think that SQF can evolve in a programming language with similar capabilites like Python etc, because BIS surely does not want to invest so many resources in doing so. An exception is UDK from Epic, which has unreal script as a propretary scripting languge...but the issue here is that it is a full-blown programming language with strong OO concepts and third-parties provide a full Visual Studio plugin with debugging capabilites and so on. Sure, SQF is nice if you wanna create missions, where you need here and there a script. But do you really wanna implement complex algorithms in that language? How do you debug it? How do you leverage open source libraries for AI and other stuff? How do you do unit testing ? How do you make real reuse libraries? Also if you go to industry standards, you can leverage from new trends, like functional languages (.NET -> F#, Java->Scala; partly Python). Think about how shall SQF evolve...there is no real library concept (apart of that new 'function module' stuff) If you add more and more commands and isolated functions, surely the 'hard core' of the modders can cope with but don't you think this will scare off new people? I can understand people, who wanna take their investments of Arma1/2 into Arma3. On the other hand, I think it is time to make a radical switch now. The Arma engine as a sandbox is very strong. Create a library for an industry-standard language with more low-level access (or make it more transparent)...I am quite sure people will create 'miracles' with such an engine.
  8. hawk66de

    scripting

    ok, did not see that...then this thread can be closed
  9. hawk66de

    scripting

    @Devs: My hope would be that the whole scripting approach changes to industry-standards like Python/Lua with an OO-library and not hundreds of scripting commands and no real debugger. Also to include own .DLLs would be awesome! Don't get me wrong: The engine is very capable and open but it would be just awesome if you take this road. If that's not possible due to time/resource constraints: An alternative might be to provide a bridge to those mentioned technologies (so calling a python script directly from arma script etc.). So then you can use ArmaScript to control the engine and Python etc. to implement complex algorithms, which are debuggable etc. Just my 2 cents.
  10. I've played with it and after a bug fixing ;) in validateParameters.sqf (the logic did not consider that the _logic was always the first parameter in the array if the method to be checked is a constructor, thus it rejected a scalar constructor parameter...) a simple example works^^. I'd have the following questions: I assume inheritance is not support in the released code, yet? I'd expect that in instance methods you also get the _logic passed so that you have access to the instance (similar to _self in python)...or do I oversee sth? Hope, you guys continue that OO-topic^^ Is that OO-approach only for the time being, until it is replaced by a built-in language construct?. If not, a tool to generate the method stubs by parsing the config, would be quite handy^^
  11. Thanks sbsmac for that create tool! Just one proposal: If a file is saved, disable ('grey out') the save icon in the toolbar :).
  12. I have seen that many delivered scripts make use of debugLog...in the wiki it is mentioned that this would be inactive in the retail version. Any reason for that...I mean why just not have an option that the log is written or is it written and I have overseen sth? :)
  13. Or if there is no time for writing a docu, perhaps just attach an existing example for self-study?
  14. not really...perhaps some experts here know how to create custom animations and how much efforts it would be?
  15. Since I want to start a new project from scratch and also want to play with machine learning algorithms, an OO framework would be quite handy. Any chance that you give a very brief overview or refer to an example how to do this :)? Is it planned to add OO as a core feature into ArmaScript? That and a real debugger would be just so cool :cool:
×