In regards to the "aion-emu" project...

This is in response to actions that have occurred between myself and the aion-emu administrators earlier. It's a pretty fair assessment of the situation. We've heard people making fun of groups writing emulators for AION in java and it's clear now that the language of choice is not the only ignorant and uninformed move these people have made:



When people can't be convinced that just because a discussion of an algorithm or design strategy is in C/C++ that it can't be useful to implement it equivalently in Java, there's no hope.

This framework is poorly built, suffers from extensive and overused OOP abstraction and data hiding where it's completely unnecessary (for instance a quick example: the Dispatcher class should not even exist, it is redundant twice over). It also uses a polling select on partitioned sets of connections. This is one of the worst methods of implementing a large scale persistently connected server like an MMO server, however the developers who are actively working on the project will not acknowledge their code is any less than perfection and therefore will not even look into more efficient means.

Not to mention: the documentation provided is poor and by no means adequate, tools are lacking, and the project suffers fr om quite a few other problems.

As such, I would not advise you continue here in any way supporting it. There are lots of people here advising the move to C++ for an emulator (which makes sense, because it's what should be done). If you want a real AION emulator, I'd suggest forking the project on SF.net and converting it to C++. It should take about 1/20th the lines of code that are here to do the same thing and it'll run about 10x more efficiently. With a well researched and tuned networking core, an open source C++ emulator could outperform the official servers.

This is the point of opensource. Contribute, and if you can't because the maintainers are cocksuckers who think Java is superior in every way to anything else out there, and are unwilling to even consider theories on networking strategies, fork the project and make it 100x better than the original so those authors are discredited and you can legitimately say "lol, i told you so, idiot."

Besides, they barely know English and aren't that bright. Do you really want to answer to them? :).

Quote:

Originally Posted by jM3 This is in response to actions that have occurred between myself and the aion-emu administrators earlier. It's a pretty fair assessment of the situation. We've heard people making fun of groups writing emulators for AION in java and it's clear now that the language of choice is not the only ignorant and uninformed move these people have made: When people can't be convinced that just because a discussion of an algorithm or design strategy is in C/C++ that it can't be useful to implement it equivalently in Java, there's no hope. This framework is poorly built, suffers from extensive and overused OOP abstraction and data hiding where it's completely unnecessary (for instance a quick example: the Dispatcher class should not even exist, it is redundant twice over). It also uses a polling select on partitioned sets of connections. This is one of the worst methods of implementing a large scale persistently connected server like an MMO server, however the developers who are actively working on the project will not acknowledge their code is any less than perfection and therefore will not even look into more efficient means. Not to mention: the documentation provided is poor and by no means adequate, tools are lacking, and the project suffers fr om quite a few other problems. As such, I would not advise you continue here in any way supporting it. There are lots of people here advising the move to C++ for an emulator (which makes sense, because it's what should be done). If you want a real AION emulator, I'd suggest forking the project on SF.net and converting it to C++. It should take about 1/20th the lines of code that are here to do the same thing and it'll run about 10x more efficiently. With a well researched and tuned networking core, an open source C++ emulator could outperform the official servers. This is the point of opensource. Contribute, and if you can't because the maintainers are cocksuckers who think Java is superior in every way to anything else out there, and are unwilling to even consider theories on networking strategies, fork the project and make it 100x better than the original so those authors are discredited and you can legitimately say "lol, i told you so, idiot." Besides, they barely know English and aren't that bright. Do you really want to answer to them? :).

While I disagree directly with your overall analysis of java vs native code(unless I'm mistaken, you implied java could never hope to equate native code), I do not know the quality of code which that project uses.

However I'd like to point out that English language skills don't define a single thing about how bright someone is.

I support your move to whatever language you wish to use, and I'm sure everyone would like to see a move like that(or to see some serious activity in the AION private server community).

Good luck!

Agreed.
I told them so on PP and we discussed a lot about it, cons and pros about java when it's comes to like you said : "a large scale persistently connected server aka a MMO server" but they still say java is better and no c++ discussion plz, lol come on.

Perhaps they were overhooked on AExtreme handling 3K player with a nice stressing feature implemented on their servers : moving and see others ppls. x_O
Now, when you will implement instance stub, massive world event, skills etc.. would love to see how 3K becomes 500 ppl lol. I didn't looked much at the networking part however I was surprised on the lack of pooling DP. BufferPool hi ?
In the end, just want to say that java is not the language to do this job and a forking is just needed.
My friend has a C++ emu up to date but closed source atm. There were some attemps during this summer but no updates. I passively continue working on my framework until someone feel up leading an emulator project for aion.

all you need to do here is to take the C/C++ Lineage Pserver against the Java server that can barelly handle 700 to 1000 players against 3k on the C/C++ server...

Also there is massive reboots from time to time to clean up the java while on the C/C++ you dont have to with such a frenquency and so on...

JAVA FAILS badly in this kinda of thing

Quote:

Originally Posted by andyd123 While I disagree directly with your overall analysis of java vs native code(unless I'm mistaken, you implied java could never hope to equate native code), I do not know the quality of code which that project uses. However I'd like to point out that English language skills don't define a single thing about how bright someone is. I support your move to whatever language you wish to use, and I'm sure everyone would like to see a move like that(or to see some serious activity in the AION private server community). Good luck!

In this type of project, Java will never perform as well as a C/C++ server could. If they developed it flawlessly by today's most advanced techniques and standards it would maybe hold ~4000-5000 people on a damn powerful system fully developed, but on C/C++ you could easily get 17-20k. The reasons are quite obvious, some optimizations in C/C++ can reduce a lot of things to wait-free and O(1) [and here that is ONE cpu cycle] whereas in Java this is simply not possible. When you want to keep concurrency and throughput at a maximum so that things like memory copies, context switches, etc aren't slowing your system down, these are things that just aren't nearly as easily done in Java. Java has its place in certain types of projects but high impact applications like this one (and 3D games for instance) where MAXIMUM performance and raw horsepower are needed.. is not one of them.

Quote:

Originally Posted by maxki Agreed. I told them so on PP and we discussed a lot about it, cons and pros about java when it's comes to like you said : "a large scale persistently connected server aka a MMO server" but they still say java is better and no c++ discussion plz, lol come on. Perhaps they were overhooked on AExtreme handling 3K player with a nice stressing feature implemented on their servers : moving and see others ppls. x_O Now, when you will implement instance stub, massive world event, skills etc.. would love to see how 3K becomes 500 ppl lol. I didn't looked much at the networking part however I was surprised on the lack of pooling DP. BufferPool hi ? In the end, just want to say that java is not the language to do this job and a forking is just needed. My friend has a C++ emu up to date but closed source atm. There were some attemps during this summer but no updates. I passively continue working on my framework until someone feel up leading an emulator project for aion.

All they ever do when you bring up anything, ANYTHING, that relates to any language other than java (even if it's an example of an algorithm that they could implement in java) they instantly refuse, and say "STOP TALKING ABOUT C++" as if they're stupid as shit and don't realize that algorithms and design approaches transcend language barriers. When their server is peaking at ~300 players and they want to know why, I'll laugh.

I must admit that I have not looked at the source at all but as an open source game engine developer I MUST agree with everything jM3 is saying in this thread.

Usually such arrogance as what jM3 is describing when it comes into suggesting algo's with c/c++ code examples (or whatever language...I like to use inline asm for certain things that could benefit from the low level implementation hence if it was me discussing a rt thread scheduling algo for example, I could have given an asm example), usually comes from the fact that said developers are actually clueless on other languages. That's probably what they've learnt and use and do not understand why it's not the proper toll for the job.

That said, JVM/interpreted/bytecode/managed_code/JIT_compiling can not outperform native code except for very simple projects. That is a common fact. The only reason someone would be willing to use java is it being cross-platform. Still, if you want that, you can achieve at least winapi/*nix compatibility if you have that design in mind (And albeit a million ifdefs:p)

Regarding "when their server can only handle 300 players and they want to know why", I doubt they'll even want to know why...if they had any idea on profiling and performance optimizations they wouldn't choose java for such a project in the first place... I mean you can't possibly expect them to profile their application when they do not even understand their chosen language's bottlenecks...

I can agree that C/C++ will run faster than java code, however to the people who think java is not suitable for this type of project i disagree on. Runescape's client and server are both written in java and run completely fine. I have seen quite a few private servers written in java that can handle thousands of players perfectly fine. The fact that C/C++ can handle more is a +, but is it really necessary to have that many people online? No games have anymore than 3k players online on each server due to getting crowded.

@namerdrisk: that linage server i bet was written by a bunch of newbies who barely know how to write efficiently in java. I've seen C servers that have been written by morons that crash often with 50+ players online.