next »

[update for those who read this topic for the first time:]

We have kinda decided to use C# with XNA, but there's a big drawback: it's not cross-platform, which is kinda standard in open-source applications. As an alternative we could use C++ (or Java, but that's slow) with Irrlicht. What are your views on this?

update for those who read this topic for the first time: Java is not slow: http://openwar.bartvanheukelom.nl/forum/index.php/topic,124.0.html

In light of this subject, I would like to add this link:
"What language was the game engine developed in?"

The link brings you to a topic on the Bioware discussions board. The game engine in question is the Dragon Age engine. Dragon Age is not a RPG game, but I doubt that matters very much. (BTW. Don't stop reading till you've seen it all.)

Personally, I think Java isn't an option. In the above thread, and a few others besides, it's described as a slow language if it concerns anything graphical.

That would make C/C++ the only other option. Working with these languages will be thougher. Also, Irrlich isn't as flexible as .XNA, which doesn't make it any more suitable for our purpose.

Seems it's flexibility against cross-playform.

Solinx

Also, Irrlich isn't as flexible as .XNA, which doesn't make it any more suitable for our purpose.

i know i said that, but i doubt i now. i don't even know why i said it anyway

edit: now i remember why, but those things only apply when it's used with .NET

Cross platform is definitely worth striving for, even at a higher cost of time and effort.  If it's within our grasp, we should strive for it. 

I would also like to say I agree with this statement wholeheartedly:

Give me any language and a good class library and you have a great combination for building systems.

Language is just a tool.  You can give programmers good tools or bad tools, but it will be the design techniques they can muster that will define the end product.

I'm also seeing that garbage collection may in fact be a bad thing.  I thought at first that garbage collection would be very good if we have a novice group of programmers, which might accidentally create a memory leak (which is DEATH on a huge scale, but small budget project like this).  However, it seems that even the best garbage collectors are just too inefficiant to run in the middle of a game, so that's not an option.

which is DEATH on a huge scale, but small budget project like this

could you elaborate on that please?

Memory leaks are hard to hunt down on a huge program, since there is largely no way to know where they're coming from.  Essentially, you have two options: the "old fashioned" way (line-by-line code reading... not practical in a program as large as this), or buying an expensive program to do it for you.  Essentially, it's the proverbial needle in a haystack...

Wouldn't it be possible to pin point it to a part of the code? I mean, if a program had no leak just the day before, you introduced it with the code you wrote that day, or triggered it at least. That should make it easier to find a leak, shouldn't it?

Solinx

The problem is when you notice the leak.  There's no assurances that you noticed it immediately after it was added.  While there might sometimes be hints, a memory leak might exist in one of your earliest pieces of code, but might have only become substantial once you compiled a large section that works together.

unfortunately that's true for any kind of bug

Well, with a bug you can always evaluate what the bug does and start you search in the general vicinity of the code that does the thing that's bugging up.  With a memory leak, the symptom is memory that is not being deallocated properly.  You have pretty well nowhere to start since there is no abstract function to tie this particular bug to.  That's why memory leaks are among the worst bugs to have.

[malloc]

With a memory leak, the symptom is memory that is not being deallocated properly.  You have pretty well nowhere to start since there is no abstract function to tie this particular bug to.

you can simply find all instances of malloc, and follow the variables until you find where they are freed (or should be)

anyway, about cross-platformness, I think we should support windows, linux and mac. other OSs aren't really possible because most libraries (which we're going to need) are for those three only

Windows, Linux, and Mac should cover 99.9% of people anyways... and those who pick others fully understand there are compatibility issues.

And in this case, just say that they can ahve linux for free

Windows, Linux, and Mac should cover 99.9% of people anyways... and those who pick others fully understand there are compatibility issues.

Is there any reason to bother with Mac, anyway? The vast majority of users are on Windows, and most of the rest are on Linux. The market on Mac for games is so small it's not worthwhile;  Probably 99% of the people on Mac are there for photo/video editing or general usage. If someone does have a (newer, Intel-CPU) Mac, they can use Boot Camp to run Windows anyway.

there isn't really a reason, but it "comes for free" when you use Java

i would love to see that done in Java. mainly cause my bachelor's project is programming in Java so then i can  join fun and work

you're most welcome