Mikkel

Whilst dwelling over a forum which became so engulfed in flame it required the fire brigade, I began to wonder why one person loves one racing sim and another so passionately despises it. No prizes here for working this out for yourself of course - there's no perfect answer. This however, led me onto a further question: What components from which games would be a perfect hybrid - one which everyone would be happy about…indeed could this be possible and would we want it.........

Read the rest of this column here...
http://www.racesimcentral.com/artic.../column10.shtml

And remember to post your thoughts and comments here !!

__________________

Mikkel Gram-Hansen || - Always second, never winning. *sigh*
Danish GPL Society || Team Leader

- Admin RaceSimCentral & SIMRACER ! :D

Marty

You are of course, absolutely correct. A commercial product will never satisfy us all, because that's not maximising profit. Why make one perfect product when you can convince the majority to buy several products that nearly do what they want?

If it's going to happen, I'm guessing it's going to have to be developed by someone who isn't controlled by marketing (potentially nK, RL or LFS? What are Papy upto now?) or when us consumers stick together and refuse to buy anything that's less than perfect. I'm just a guilty as everyone else here - I play a Nascar game about once a year, and I consider N2003 to be worthy of a patch to N4, yet I own all three of Papy's Nascar sims.
Complete rubbish, in my own humble opinion Reading marketing hype for twelve months prior to a games release is anything but fun or interesting. Besides, I can think of far more interesting things to talk about. In the 'old days' I would have agreed, but now there's so much mediocrity about it's depressing (it could of course just be that I'm now older & more synical / depressed).

I'm hoping that piracy becomes so wide spread that it becomes impossible to make profit from video games / movies / music etc. Maybe then we'll just be left with people who create this stuff for the right reasons. We must be due for a miracle sometime soon mustn't we?

stenyak

Reading that, one thing has come to my mind: the linux kernel.
For those who don't know, it's an operating system written by a LOT of people all over the world, with no economical profit. Working on it just because they like doing it, because they want to provide an alternative to microsoft programs... whatever.

What if there was something similar but with a racing game? I know, there wouldn't be _so_ many people working on it, but.. lfs is being created by just 3 persons. Also, look at racer... just 1 person. NetKar too. Look at RacingLegends...

Well, now imagine there was... let's say, 25 persons all over the world working in the same racing sim. Working in a new 'game', in THE Racing Simulator.
Some of them working in different parts of the physics engine, some of them modelling tracks, some of them working in cars, some of them working in sounds, others working in the interface... and then a couple of thousand testers. And anyone who could help would help. And of course, there would be someone / somepeople who coordinate the work.


Do you think it's really impossible to have the perfect racing sim?

Don't you think it's time to start a project together?

__________________
Petition for a Software Patent Free Europe

MOtorSPort - the Racing Enthusiasts Vehicle Simulator

Marty

Far from it.. but as you suggested, we are going to have to create it ourselves.
Absolutely. The huge number of mods and "free" sims available is proof enough that we have the talent.

The hardest part will be the core physics engine, but as you pointed out this can be handled by one person (as is the case with pretty much every other racing sim). There are people now who I know are capable of fulfilling this position but they must want to work on a GPL style project and be prepared to handle (and put with) bug reports, feature requests and (potentially) lots of abuse.

The first steps will be the hardest - once some structure is in place and we have gained some interest from the current popular mod teams or developers it should be downhill all the way.

I imagine the structure would be along the lines of Emacs - as little source code as possible with a good scripting engine. The scripting language would be used to define everything from the menu interface to running a championship season on a multiplayer server. Doing it this way would allow a larger number of people to tinker with as much as possible and submit updates & bug fixes. I think this method would have a higher chance of satisfying everybody since it shouldn't be too difficult for anyone to learn how to change whatever they don't like.

What we need to move away from in my opinion is the concept of mods. The "now we load this carset with these physics and we can race in this championship" style. What we should have is a huge number of cars and a huge number of tracks. The interface should then let us say "I want to start a championship season running these tracks in this order, and I want to allow all car types without wings to take part" or whatever.

Personally I would love to get involved in a project like this (I have lots of experience writing C, a little knowledge of C++ and a fair amount of experience with various interpreted languages) but unfortunately I know very little about implementing car physics, graphics, sound or networking. I don't think I'm ever going to be fully satisfied with a commercial sim so here's hoping some nutcase steps up to the challenge and gets the ball rolling...

stenyak

In my opinion, the physics engine is [should be] the most important part of a race simulator. It will not be supposed to be a fun game, but a realistic simulator. And racesim fans will therefore find it fun. It will not be supposed to be a game appealing to lots of people (not like typical commercial games).

So, if possible, it would be better to have more people working on the physics engine. At least 2 persons. Keep in mind that, being this a spare-time project, we need more people to make it playable (with no important bugs, etc.) in a reasonable time. It will allow continued work, and probably better code than if there was just one person.
I would like to say that from my point of view, this project should not be attached to any particular developer or mod team. You know, from simracers, for simracers. I expect the project to be evolving, with new developers joining while others leaving.
I haven't read about the code of emacs, but the idea should be what you explain: have a flexible source code that allows to easily add new features to the project. Why should we need mod[ification]s (as seen in commercial games) when we have the source code?
This discussion has also been present in the development of other free programs, such as eMule, and as a result, i conclude that the only "moddable" aspect of the project will be the [easy and semi-automatic] addition of new tracks/cars (+championships&similar things?). Something like racer: copy a folder and there you have it.
It's good to have an internally powerful/versatile structure, while keeping a simple look/use for the end-users (for the most commonly user-modified features of the project).
I also would really love to join a project like this. My knowledge of C is not really good (i've just started programming in C three weeks ago), but i have experience with some other languages and i can learn quickly if needed. I have done some basic networking and multithread code in C under linux (haven't done anything that complex under windows yet). However i guess networking code will be left aside until we have made the basics of the project. I have almost no experience of gfx under linux (i managed to show a triangle with opengl in a window under linux, but didn't do much more than that), mainly because i can't find basic (and easy to understand) documentation of it on the net (or at least, examples that work).
Also, i know a guy who has studied car physics for about 6 months, so he could be of help. In fact, we were about to start making a car game.


I suggest we start thinking more seriously about all these things, maybe starting a new thread or something else (if it is to be a open project, good communication is needed).
Let's start this project now!

__________________
Petition for a Software Patent Free Europe

MOtorSPort - the Racing Enthusiasts Vehicle Simulator

Marty

Hi Stenyak,
Yes yes, of course. Don't forget that this means it will be VERY difficult to write!
Absolutely, what I really meant was one person will probably have overall control and make sure only good code goes into the main build tree.
Again, yes. My comments were more about raising interest in "the" project. People who have worked on existing mods or their own engine will have experience and expertise which are what a project like this would need. The failure or success of a project like this would depend on the community pulling together, including (at least some of) the existing mod teams / developers IMO.
I haven't looked at the code, but yes the structure of Racer is heading in the right direction I think. Being able to use Racer (for example) as a starting point would really help a project like this, if Ruud decided he didn't mind a bunch of "hackers" like us attacking his code
Experience is important, but I'd say it's more important to have the right mindset (is that a word?) for it. I've seen terrible code written by supposedly "experts" who've been doing it for years and some good code written by newbies using Notepad (ok, not quite but you know what I mean).

To implement car physics, it's of course more important to understand car physics than it is to write C or C++. I could probably write the code if someone could give me all the right equations!

Networking I'm guessing will be the next hardest area, here experience definitely counts.
Yep, there does seem to be a bit of the "why do you need documentation, you've got the source code?" type attitude. But I think that's a trade off from allowing huge numbers of developers to contribute (and a fast development cycle). I'm actually more interested in *BSD than Linux as their ethos (or whatever you call it) is more inline with my own thinking.
Good.. strap him down to his desk and don't let him leave until you see some results

Please post here if you start another thread (or if there's existing ones I should read).

Regards,


Mart.

kunos

Hi all,

Interesting thread going on here...

I don't think the project is impossible but... I have to point out that it is not as "simple" as writing the linux kernel or emacs.
I'm going to explain what I mean... things like a text editor and even a complex thing like OS kernel are still part of the domain of electronic and software common knowledge...
Why you see so many free window managers, email applications, webservers and so on but NO (apart from some exceptions) free games, high end 3D applications, high end sound applications?
Simple, because they need something more than a "common" software knowledge.. being a good coder is not enough to write a simulation.. you really need to deeply understand what you're trying to simulate...

Having a team of 2 coders doesnt cut the development time in half.. just the opposite, it could actually make that time longer.. because they will need planning, coordinations, modularity... many things you can do on "autopilot" when you're on your own...

as for myself.. I find the project really challenging.. sadly, as you might notice, after my move to JP I hardly have the time to fix things in nK, sure I can't join something requiring such a level of involvement...
but I'm ready to collaborate anyway... being it as an outside helping hand, and with code.. things like multiplayer parts in nK are quite modular.. I would be happy to share the source code with the devs of this new project (as long they don't become public, for obvious reasons.. cheaters)...

best luck anyway..

__________________
Stefano "kunos" Casillo
Ayu Fan
http://www.kunos-bside.net

Marty

Thanks Kunos

stenyak

Hi Marty! I have been thinking that we should start with a basic but powerful physics engine. Don't how to explain this, let's see an example:
-In the beginning of the project, the wheels spin at the speed of their side of the axis. The speed of the axis is the speed before the differential, equal to the speed of the end of the gearbox. The speed of the gearbox is equal to the speed of the crankshaft (is that the word in english? i mean the value of the rpm-meter). The speed of the crankshaft is proportional to the speed of the pistons (...the thingies of the cylinders). And this is proportional to the fuel given, equal to the value of the input (gas pedal, keyboard 'up arrow', whatever).
This is a very basic and very unrealistic physics engine, probably unplayable and with missing things and maybe some others wrong (it's just an example).
-The next thing to do with the physics engine would be, for example, changing the part of the "final gearbox speed equal to crankshaft" to this other "final gearbox equal to crankshaft times the current gear ratio" (or whatever math. formula is to be used).
-After that, we can change "final gearbox equal to crankshaft times the ratio" to this "final gearbox equal to initial gearbox times the ratio", and then add a "initial gearbox speed equal to speed of crankshaft divided by clutch value (third pedal, keyboard 'c' key,...)".
-All like this (modofying and adding new parts) until... until we want to, or... until we have a perfect simulator (well, a really-close-to-perfect simulator ).

This way, we can start with a speedOfWheel=inputOfGasPedal, then make the engine better by adding more parts or changing exisintg parts with more accurate formulas/data/whatever.
And also, we can have different people working on different parts of the physics engine _at the same time_ (similar to people who use libraries whithout knowing the internals of them).
Just an idea.

I agree. We _need_ good coordination between the coders if we want this project to evolve. There must be some coordinator[s].
Still haven't looked at the code of racer, but since it is available (some old version, but that's enough for our purposes), we can look at a good starting example.
As i said in my first post, not everybody need to be coders. We should have some people who know how to code, but also people who know modelling, who know about physics (the guy i've talked about), etc. That's why a think it's better to have more than one person working on the physics engine: if they are specialized in a thing, they will probably do it better than having only one person working on all (writing code, reading about the maths, etc.). However IMO it's necessary to know a bit of what the others do (imagine we (coders) didn't know... what a clutch is).
I don't know much about BSD. Will the linux source-code compile under BSD easily?
I think we need to make the project executable in the platforms sim-racers mostly use. That means Windows should be supported (not that i like it, but would help to attract more possible contributors to the project). And also Linux. I think those are the two most important o.s. nowadays. However i don't know how many people use BSD, MacOS, etc. We should think more about how to approach this. IMO trying to make the game compilable under several (too many) platforms will make it harder to enjoy the development of the project itself (i mean the real interesting things: physics, gfx...).
It's obvious we need some place to share ideas, code and... well that kind of things. The rsc forums are a good place to share ideas. But about all the coding, maybe sourceforge? And bug reporting, feature requests,etc?. I don't have a clear idea about how should all these be done.

Hi!
If the coder and the physics guy can exchange ideas with no problem, then the coder doesn't really need to have _deep_ knowledge about physics. Well i think (and hope) so.
That's true. I first thought of creating a car game by myself. I usually work better alone because otherwise i have to explain what i do to the other, sometimes i even have to explain basic things the other should know about, etc.
But beginning to think how to do the car game, i realized i would need some help with the physics. Then i realized i had no idea about creating a gfx render engine... Even if there's some time lost in communicating with the rest of the coders, it's better for them to know about what you do.

I think it's better to have the project with several people working on the same areas (at least in really _similar areas_). That way the work will not be interrupted if the 'main' coder can't keep working on it, or if he decides he's tired of it, or whatever. As i said i expect this to be a dynamic project, not dependant on one or two coders, but on the community.
Mmm hadn't thought about that. Cheating in multiplayer, given the source code is available...
I always thought there was no uncrackable code (via reverse engineering, not force attacks), but then i read about the autentification system used in linux (coding the pass based on the pass itself).

Maybe there's a way to make cheating difficult even with the code available. Anyway, if the source code is versatile, then the dev team (the coders) could work in (share between themselves) the multiplayer secret-code, giving the community (users and such) access to the _binaries_ of it. Of course, this would require the coders not to reveal the source code of the multiplayer part. Maybe the community is loyal and can be trusted with no need for spending too much time on anti-cheat code.
Anyway, there's plenty of time to think about it. First we need to do many other things.
Thanks for your support!


edit: As a reminder (phew long post above! ) : now we must focus on finding a way to coordinate the efforts. Please write suggestions about what i've said (rsc forums, sourceforge,...).

__________________
Petition for a Software Patent Free Europe

MOtorSPort - the Racing Enthusiasts Vehicle Simulator

Marty

Isn't that an oxymoron
Simple is good (to begin with), so yes that's the right approach. From what I've read modelling most "parts" of a car isn't too difficult and I think it wouldn't be too long before we had something that felt like a reasonable approximation of driving a car. Aerodynamics I guess wouldn't be too complicated either as I think it's more "common knowledge" as Kunos put it.

However, the problem is going to get proportionally (exponentially?) harder the closer we get to reality. The big area of black magic would appear to be the tires.. the tire model is going to be the killer I think, where someone with experience and expertise will be required.

I imagine beta testers will be the major contributors here, but it's not going to be easy as I expect most testers are going to be saying things like "it feels like the flywheel is sticking when I'm off throttle approaching this apex changing from 3rd to 2nd and the slip angle is sub-optimal for the corner" or whatever. Someone has to interpret that and understand what it means in terms of formulas.

There are of course many other issues - I have no idea how collision detection works for example. Oh, and of course I have a full time job eating up a large chunk of my time and killing my enthusiasm for writing MORE software when I get home...
Yes.
Indeed. One thing I've always thought is that the first thing that should be modelled is a car body without any wheels, as if implementing a flight sim. The body should then be dropped from different heights / angles and pushed along with different forces. Only when this is realistic should the thing have wheels put on...
Don't forget it's copyrighted to Ruud, so none of the code can be "copied". But yes, it would be a good source (ha ha!) of information for us newbies.
Pah, what the hell... I think we should just beat him with a stick until he agrees to let us use it as a starting point
Something to do with straws right?
I don't know much about cars, but I'm still here
BSD has support for Linux binaries, and yes (I believe) most Linux code will compile under BSD with little problem.
I'm only just starting to look at BSD so I don't know for sure though, but I think Linux & Windows are the only important OS's that we need to worry about (MacOS is based on BSD so if Linux is supported pretty much every other OS should be an easy port anyway).
Not completely relevant, but if you haven't seen it Cygwin is an excellent environment that would allow the same build system and tools (eg Perl, a personal favourite) to be used. Obviously the main code should be built on Windows with MVC (not gcc) for best performance though.
We need more than that... we need a miracle!
Well not quite, but I don't think this project is going to go anywhere unless someone with far greater knowledge than both of us is interested.
SourceForge does have bug reporting & tracking, which can also be used for feature requests. Never used it myself though...
There are existing projects such as OpenAL, OpenDE and SDL which may be of use.
I thought Kunos meant people stealing the code, not cheating in multiplayer...
Anyway, Linux (open) has far better security than Windows (closed) so that shouldn't be a problem. In fact being open source can help as it means any exploits are generally picked up far quicker.

BTW, it wouldn't bother me if this wasn't an open source project. As I mentioned earlier, I think much of the "code" most people want to play with can be pushed into scripts anyway. If this method would allow us to use say a network module from someone else then all the better.

stenyak

I expect to learn while doing this project, and i don't expect it to be... mmm 'finished' in X months or years. I'ts not gonna be sold in crhistmas or whatever, so i can spend time learning.


That's an approach i hadn't thought about. I thought we would first make a basic 2D physics engine, then begin with the third axis.
That's what i meant: using the code to learn, not to copy it into our project.
I am determined to begin a car game/simulator. But, after reading this thread, i thought i could not only do a game for myself (for my tastes), but for the comunity. And therefore, let the community do coding and other things (not just me).
That's my idea. If nobody wants/can join the project, i will start with no help. Probably the first thing i'll do is buy a book on using opengl under linux
Interesting. I was thinking to develop our own gfx/sfx/physics engine, but since there's a gfx and sfx one already available, maybe we should use them. However i'm not sure about using ODE. Don't fully understand if ODE is already a complete physics engine, or if it will be just a base for our physics engine, or if it will be modified by us in order to make it [feel] more realistic when using cars.

Mmm... no opensource, no sourceforge What do you suggest? I don't think opensource will harm or anything.

__________________
Petition for a Software Patent Free Europe

MOtorSPort - the Racing Enthusiasts Vehicle Simulator

Mikkel

wow guys please keep this discussion going, it's highly interesting !

__________________

Mikkel Gram-Hansen || - Always second, never winning. *sigh*
Danish GPL Society || Team Leader

- Admin RaceSimCentral & SIMRACER ! :D

Marty

On a side note, went to see Ben Harper last night... damn he's good

Anyway, ...
Good, and I will help if I can. But.. for me this is not a learning exercise or about creating yet another car sim. It's about (the idea at least of) making the car sim that will spell the demise of all others, monopolise the genre and destroy any hope anyone may have of making a commercial racing sim thereafter.

I'm therefore only interested if there is at least some probability of bringing the race sim community together on this and making something at least on par with the combined forces of LFS, GPL, netKar and NK2003.

Personally I wouldn't start from scratch - maybe RARS. It's usually much easier to think about what you want to achieve by modifying something that exists than starting over. Plus most of the really dull ground work is done.

As we've already discussed though, the success of this depends on the community and the fact that this thread has been viewed 130 times (at the time of this post) and there's only two of discussing the idea suggests to me it's not going to work.

Ok, some ideas and suggestions:

* Get the interest of people such as Brian Beckman or even Maxx. i.e. people with a technical background and real life racing experience.
* Get the GPLEA team to finish the '67 cars without the need for any of Papys original work so they can be included in the new sim (would help to gain interest I think and allow many people to directly compare the physics of the new sim with GPL).
* Give Ruud, the LFS, RL and RBR teams etc a prod and see if they're interested in something like this (I'm not adverse to the idea of the perfect sim being a commercial project, and note it IS possible for something to be open source and make profit from it).
* Start a website calling it something like the Racing Open Standards Group or whatever, (rosg.racesimcentral.com ) using a name people will remember and get the ball rolling. "If you build it, he will come."
* I'm trying to think if it would be possible to get any real racing teams involved in making an "America's Army" for the race sim world ...
Thinking about it a bit more, I think open source would be appropriate, at least for the physics. This will allow the code to be scrutinised by many more eyes.

stenyak

I said that because i know i know nothing: i'm already beginning to learn opengl under linux, after that i'll have to learn more about car-physics, etc. That's the idea in my sentence: i want to do the perfect racesim, but in order to acomplish that, i must first learn many things.
mmm that's a problem in my personality: i always want to do things myself. Well, at least my personality suggests me not to do the project in assembly .
But the problem is still there: i haven't made any program similar to this project; this is a huge project! As i said, if i am the only one interested in [doing] it, then it will take a lot of time (probably several years) to get to a 'finished' state, no matter if i start from scratch or not. I prefer to know what does everything do, not just using a 'black box' that magically makes some cars appear on screen.
I don't think people usually read threads on the "Site Articles" sub-forum. And if anybody somehow ends here, usually he will not bother to read those _big_ posts above.
If this thread was stickied into all sub-forums, i guess the number of replies would change. Anyway, we don't need people posting a lot of "this is a cool idea" and such (at least in this thread). We already know that . What we need is to start a project in sourceforge (do you agree?), do some basic code, then post a notice in these forums (any place where people can easily see it), then let people put their feature requests, their "this's cool", and their applications (is that the word in english?) to join the dev team. Better if they do the latter.
This should be done once we have some basic code to use the cars with. But the idea of comparing both physics engines is good. However, if we want people not laughing at us, we will need more than "some basic code" before we compare the behaviour of those cars in the project.
What does 'prod' mean? (sorry for my poor english)
Do you mean, telling them if they want to develop this project?
Fully agree. However, i don't want extra work creating the web (yet). Let's first make the sourceforge thing, and after some time, if everything goes as supposed, then we can create a web site. What i mean is that, IMHO, we better don't make people be interested in possible vapourware.
Yes, i'm tired of talking about "the project" or "the race simulator"
I haven't played America's Army, so i don't know what you mean. Please, explain it


I get this conclusion:

• First we need to choose a name for the project.
• After we have a name, we create a sourceforge project.
• Then we do some basic code. As i said, being only me (+you? + possibly, but not probably, some other people), IMO this project is likely to end as vapourware. I don't want that to happen, so i'll try my best in order to have some basic code. That's the first target, and from my point of view, it's a really big wall to climb.
  Right now i was doing some cubes in opengl. Can't even think on how i'll do to read car data (what data format?) from hdd and display it on screen. Not to talk about sounds,etc. Maybe it's easy to do and there will be no problems, but right now i just got no idea about it, so i see it as a big wall.
• Next, create a web site explaning what we have done, what we aim for, and let people know about the sourceforge project (feature requests, dev aplications, etc.) via the web site and via these forums.
  Of course, anyone who want will be able to join the dev team at any time. What i don't want is to have a lot of people creating cars and tracks for a program that doesn't have a line of code writen.
• ... next things to do are still unknown

__________________
Petition for a Software Patent Free Europe

MOtorSPort - the Racing Enthusiasts Vehicle Simulator

kunos

I agree with Marty and I dont think Stenyak's approach is correct.
The original idea of the thread was to create a combine effort that could put together someone who already proved his/her (I wish there was an "her" in racing sim developing lol) in other products..

Stenyak's idea looks to me like the other 5-6 indipendent racing sims popping up around, it looks like every racing sim website has "its own" racing sim under development.. it is kind of funny.. but I understand the feeling because I went "oh, well.. nevermind I'll do it by myself!" too...

But Stenyak's plan looks like a recipe for failure really.. 2 points in my opinion are a big "no no":

1) You say OpenGL under Linux, and it is a BAD idea... for a project like this you need to MAXIMISE time usage and make it as simple as possible for the user to try your game. From the time point of view, Linux game development is virtually non existent.. every racing sim out there is for Windows, 90% of them are DX based.. the reason is: because it is a better environment to develop.. it is easier to find infos around... and when you need integration with material coming from other software you can run it on your machine (3D studio, photoshop, sound editor ecc..)..
Plus the user.. Linux is a pain in the ass.. do you really want to force someone to install an OS to use your sim?

2) The idea of "starting with a 2D sim then evolving it into 3D" is suicidal.. .pure and simple. I can assure you, it will not work.. I did the same mistake 4-5 years ago. The way to do it is to build a basic physic engine you're happy with (3D of course) THEN think about cars and racing, not the opposite.
You can't fake physics.. it is not a process of make things more and more complex as you go... it's the opposite.. it is creating a basic environment that you use to build upon.. in other words you go from the complex bottom part up to the racing simulation.. not the opposite.. from the racing sim down to the difficult physics parts..
The more code you create, the more difficult it will be to modify it. I strongly recomend you to abandon this course.

all of the above is, of course, IMO...

EDIT: Just as an example of what I am saying.. look what Chris West is doing... he's not putting together a simple 2D physics engine to make screenshots and then develop the rest.. he's doing the BASIC physics part (with Gregor) first... then he will move into the "game" part..

__________________
Stefano "kunos" Casillo
Ayu Fan
http://www.kunos-bside.net

stenyak

Wait a moment. I haven't said to do the project exclusively for linux. I said the project should be made for both windows and linux. The thing is, i have started learning opengl under linux because that's what i usually use. Opengl can _also_ be used under Windows. So i don't see the problem (or am i missing something?).
Also, i think that more and more people will begin using linux. It's just a matter of time. Right now linux can be difficult to install and configure, but IMO in 2 or 3 years things will have changed, and it will be easier for the users.
Wait again . I just stated that the approach of "doing a 3d engine then put the wheels to the car" was new to me. I stated that, from my inexistent experience in creating car games, i had thought that doing a 2d then add the 3d dimension might be a good approach. But reading your comments, it's obviously wrong.
Maybe i should emphasize that, whatever i say here is not to be taken as "truth" or as "what will be done". They're just ideas, suggestions. After all, the idea of this project is from simracers for simracers; I don't want it to be just _my_ project, otherwise i wouldn't have said anything in this thread.
Maybe the problem is that i'm the only one who is actually going to code, at least in the first steps of the project .
I agree with you on that point. That's why i think we should think about things before we begin coding them.
The thing about doing a 2d engine then a 3d engine, maybe i should have explained more my idea:
what i say is that, in order to have something working, it seemed for my easier to do a basic 2d engine (and by 'basic' i mean speedOfCar=gasPedalInput;angleOfCar+=steeringWheelIn put) in order to have something working. After i check there's actually something in the screen, doing what it's meant to do, then i can start doing the real 3d engine. The 2d engine is just a way to climb the big wall i talked about. It just seemed to me (and still seems ) to difficult to begin directly with collision detection and physics engine under a 3d environment (if you want the car to lay on the ground without wheels, collision detections is needed... am i wrong?).

Anyway, i thank you kunos for your comments, it's good to have somebody giving advice

__________________
Petition for a Software Patent Free Europe

MOtorSPort - the Racing Enthusiasts Vehicle Simulator

stenyak

Hello everybody! (is anybody still here?)

I have been doing some things these days, from learning some generic opengl things to learning some basic sdl.
I have also been creating some basic code, just as a test (or maybe the beginning of the project ). However i have some questions that, hopefully, somebody can answer. (BTW: the code is supposed to be compilable under windows and linux; haven't tested it under windows though)

-Firs of all, what useful places do you think i should visit (irc channels, newsgroups, websites) in order to get answers for these (see below) kind of things?

-Should i use several threads for every engine (gfx, snd, physics, input and net engines)(i call them "engines", dunno if that's correct), or is it better to make separate functions and call them from a main loop (iteratively, not in paralell)?
I'm making every engine independent from the others, but obviously they share some specific information with others. The relationships are these:
·the gfx engine will read data from the phx (=physics), input (showing the input in screen like LFS) and net engine (chat messages and such).
·the snd engine will read data from the phx engine.
·the phx engine will read data from input and net engine (in order to know the position of other cars in multiplayer).
·the input engine will read data from phx engine (force feedback)
·the net engine will read data from input (chatting...) and phx engine (in order to send our own car's position, etc.).
The engines can write their own data and read (not write) any other engine's data (if allowed by the relationship above).
I don't know if i should create another engine, the 'game engine', which manages the intro/outro, menus, and the rest of the engines (synchronisation, etc.).


-And the last question: any suggestions for the name? (i need one in order to create the sourceforge project and have a place where to discuss this and other things). I suggest some:
TRS - The Race Simulator
URC - Ultimate Race Simulator
ORS - Open Race Simulator
ODS - Open Driving Simulator (it's not all about racing, right? we can also watch the landscape...)
NJARC - Not Just Another Race Simulator
...maybe they're not good names, but i don't have any better ideas

__________________
Petition for a Software Patent Free Europe

MOtorSPort - the Racing Enthusiasts Vehicle Simulator

Mikkel

I'm still here, and I'm still interested

__________________

Mikkel Gram-Hansen || - Always second, never winning. *sigh*
Danish GPL Society || Team Leader

- Admin RaceSimCentral & SIMRACER ! :D

Marty

I've been informed all you should need is one thread for graphics rendering and another for the main app (everything else). This will prevent the rendering (which will vary in duration from frame to frame) from interfering with the physics etc.
World Sports Cars
Open Driving Legends (OpenDL)
Fake Driver (cos it ain't real, it's a sim innit?!)
Rubber Soup (or anything completely unrelated to what it is)
Bland Driven Auto: Open City

Mikkel - I'd be interested if you have any comments / thoughts / suggestions about this thread, as a veteren / administrative RSC member

Mikkel

Will think and see if I can add anything usefull even if this is not up my usual alley, I'm a boring administrator/politician not a gamebuilder... but the name is easy

"The RSC"
Project RSC
The RSC Driving Experience
Community Driven
Drive
The Experiment
Racing Incident
Project Apex
The RaceSim Project
The Collective Effort

Naaaahh grasping at straws here...
I'd usually make a thread asking peoples opinion and making it into a contest.. best proposal for a name wins a small prize.

I don't have any real cool names bouncing in my head right now, but I would love for it to become a RSC pet project.. something we devote a lot of time on following and helping as best we can. I'd love to follow the process as much as the actual product, might be an extremely interesting lesson in teamwork and game development.

__________________

Mikkel Gram-Hansen || - Always second, never winning. *sigh*
Danish GPL Society || Team Leader

- Admin RaceSimCentral & SIMRACER ! :D

stenyak

But the sound engine can also interfere with physics... if we are gonna use threads, wouldn't it be better to use it with all the engines? or at least with the most cpu hungry (do you understand what i mean) ones? If it's going to be a sim, the sounds will also be rendered in real time (similar to what lfs does), so the snd engine can eat a lot of cpu.

Anyway, i have already made a gfx and phx engine (basic ones), and they work iteratively (not with threads), although they are independent (well it seems to work).

This is the code for the engines management (all the code i post is opensource, unless i say otherwise). The code gives priority to the physics engine; the [max] speed of the gfx and physics engine can be specified in steps per second (or fps): As you see, i'm still trying to do the basic programs loop. I will move onto the next thing when i know exactly how to do this thing.

In my computer, i have been able to run this code at about 485fps (gfx engine) and 960fps (phx engine; maybe it's better to call it Hz instead of fps). Or a steady 36gfx fps and 975hz for the phx engine. Well, this is just with _very_ simple engines (it's just three 2d images moving in the screen, and a very simple 3d object (8 polygons) also moving in the screen, under opengl. I have a AMD1000, plenty of RAM, GeForce2GTS (agp2x), and the code has been run under linux.

There's a problem: the SDL tick counter is not very precise when using small quantities of time. I have read it's possible to get a precision of 1ms under linux, and 2ms under windows. That means the physics engine will run at a max. rate of 500Hz (in order to have compatibility with windows. otherwise, it would be 1000Hz). I have read that NetKar's phx engine runs at 333Hz. If we want to make this the definitive sim, and if therefore we need to do calculations of engine parts in real time, i guess this is not gonna be enough (think of an engine running at just 3000rpm, that would require _at least_ a 3000Hz phx engine, probably about 12000Hz for the rotatory movement of parts, etc.). We will have to simplify some formulas. Or we'll have to search for another portable library that let's us more than 1ms precision. However i'm afraid that this limitation has to do with the o.s., and not with the library (not sure about this; do you have any idea of this, Kunos?). That's something i don't like about computers: they'll never be fast enough... for this project .
What better place to have a project like this? I hope i don't fail in my attempt...


About the name: doing a poll is a good idea. It would help promote the project, but i would feel under pressure. And if someone offers to help.... well my code is still just testing (i'm doing some code in order to see what can be done and how)... so... well... the thing is, i'm still not prepared to start the project, i need some days (probably weeks, a month or so) before i can finally start working in the project's real code. If we make an announcement right now, some people might want to code, and since there's still no real/useful code, they would start from scratch... and i would be left aside, still learning things... and i would like to be doing something in this project.... i don't know if i have explainned well...

Post your thoughts about all these things.

__________________
Petition for a Software Patent Free Europe

MOtorSPort - the Racing Enthusiasts Vehicle Simulator

Marty

Well, that depends. Taking sound as an example, I've had a (very) quick look at the SDL mixer API and it would appear Mix_PlayChannel is going to be a common function that would be called during the game. This function appears to take very little time to execute so it should be safe to call from anywhere. The actual process of playing the sound clip may take CPU time, but that will be handled internally by SDL in this case.

It's therefore really more a case of whether your calling asynchronous (non-blocking) functions or synchronous (blocking) functions.

Mix_OpenAudio and Mix_LoadWAV for example may take some time to execute, but that's ok because you'll probably only call them before playing the game, not during.
You have to think about each "engine" in turn and whether you wan't (or need) it to run as a separate process / thread, which will depend on how it interacts with other threads. Remember, you only have so many CPU cycles to work with - you're not gaining processing power by adding more threads (unless you have multiple CPUs of course).

Think about networking and the physics for example - one 'tick' of the physics engine will take the drivers and other car positions as input, do it's math and output the drivers new car position. There's no point in gathering more network data while the physics are calculating since we can't use that data until the next tick starts. Logically it would therefore seem to me that a separate process to handle network traffic isn't required.

If we need to call a function that waits until it receives some network data then of course we'll need another process for that. But, I imagine this will be handled asynchronously by events / callbacks or whatever. Your event / callback may take some time to execute, but that'll be running in the context of whatever process SDL called it with.

Hmm... I hope that makes sense!

There might be a timer function in SDL that will call a function you specify at a given rate, which may be more or less accurate.
The benefit of this method would also be that if all the work in your function is completed well within the tick period then it won't spin wasting CPU time, waiting for TickCount to become a big enough value in order to do something again. I'm only guessing here though...

The RL approach This would seem like a waste of processing power to me. Sure, make a complete engine model and see what torque etc you get at different rpms. Surely you can then just use an equation or lookup table to give the same figures in real time?
I think you have to make a decision as to whether you want to write your own racing simulator or a community one.
If you do want it to be a community project (like me) then your going to have to give way to other people taking control and stop thinking "this is my design" or "this is my C function". You can't get upset when someone rips out your 1000 line masterpiece and replaces it with their 5 line alternative that works better and runs faster.

kunos

dont use threads. they slow down the app and introduce a huge number of complications.
Sound engines are by nature multithreaded... and that's all you need really.. but make sure you have a system in place to "simulate" threads within your app.. so you can decide which frequency you want every subsystem to work..

__________________
Stefano "kunos" Casillo
Ayu Fan
http://www.kunos-bside.net

Mikkel

Check the announcement on the front of RSC.. http://www.racesimcentral.com and let me know if I said anything wrong

Once we have a name I'll be more than happy to open a new forum for this project to live in.

__________________

Mikkel Gram-Hansen || - Always second, never winning. *sigh*
Danish GPL Society || Team Leader

- Admin RaceSimCentral & SIMRACER ! :D