Forum Replies Created
If I get a spare few minutes on my desktop….. Xgl will be mine.
I daren’t do it on the laptop though, ATi drivers are very hard to get stable, and the last thing I want to do is have to mess around with X.
There’s a nice movie linked from the gentoo howto showing most of the features 🙂
Shame it’s taken so long, Macs in theory should already have all these features, windows vista will likely have them. All operating systems are moving towards requires a graphics card with good 3d accelleration. (GL for linux/Mac, presumably directX for m$).April 7, 2006 at 6:25 pm in reply to: I’m looking for some fast and free proxychecker software #28096
Strongly suspect it’s just bulletin board spam, no one can spell that badly surely ?
Strongly suspect the software it’s flogging is spyware.
VBO = vertex buffer object.
It’s a GL extension.
This is often buggy as hell with ATI graphics cards, though I’ve never had a problem with it on nVidia cards.
Maybe it would be nice to default to turning this off for ATI cards ?
Last time I purchased a graphics card, they kept trying to tell me just how much I wanted an ATI card….. I hate the ATI cards / drivers, they’re not quite as stable as the nVidia drivers.
(Yes I’m still alive)
Err, good free games I’ve come accross from time to time:
Frozen Bubble A clone of bubble bobble (available for windows, linux, and I believe there is a port to mobile phones running Simbian OS)
Neverball An extremely good puzzle game involving rolling a ball around a table.
Oolite An Elite clone, sticks very closely to the original elite, but highly customisable with many addon modules available adding new npc and player ships.
They’re the ones which come to mind immediately.
I should probably include a MUD I’ve played from time to time for years. (A MUD is an online text based role playing game for those that don’t know). Elephant MUD – http://www.elephant.org
If I were to recommend any of the above in particular, I’d probably go with oolite.
Funily enough I did use the Mersenne Twister algorithm in an earilier version of this class (CVS probably has it). But found it give problems on different architectures, so I gave up and did it a way that was garenteed to work.
Peeked in CVS, writing crossplatform code is a bitch, I have to say I’ve given up on my own project. (I made the mistake of using system network calls, and then trying to get it all working in windows, I did get it all working in windows but it looked horrible. Instead I should have used somebody elses library and let them deal with the horribleness).
rand() just sucks for generating random numbers, I think I ended up using one of the RNGs in opensll.
@The Funkinator wrote:
]It should be possible (using a seeded pseudo random number generator for the code which generates the landscape) just to pass the seed, 4 or 8 bytes would quite easily suffice for this. (as opposed to the 256 * 256 * 4 bytes requires for an entire map)
It does do it this way. But it also needs to send a compressed diff on top of this to any clients that connect after the landscape changes (due to damage).
Ah, sorry, I missed the seed() function in RandomGenerator.cpp (I was too busy being horrified by the code 🙂 )
If your interested I’ve attached a replacement random number generator based on http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html
It seems to be a good algorithm, though this would break compatability with v38 servers, so you might want to hold off introducing this.
Cheers for the ideas, I look forward to the errosion code 😉
Did I volunteer ? 🙂
I’ve done some research into this, It looks fairly unlikely that I’ll be able to implement something to apply erosion to a map, apparently you need to do something similar to how the weather prediction computers work, and I don’t think people will want to wait half an hour for the next round to start.
On the other hand it should be possible to generate something looking like a watercourse by using a L-system fractal to generate the path of the river, and then extrapolating the height of the land from this, and I don’t think you’ll need a computer the size of my living room to run it 🙂
I’ll say a few words on the lagging disconnects thing
(this might get technical)
- Client computer crashes completely
- Scorched3d process is killed via taskmanager (or eqivalent)
I don’t thing this will cause it, because the operating system should close any sockets the process had open.
- Broken NAT on router between client and server.
I’ve seen this before, and it’s highly likely as some of the NAT implementations on routers are abominable. (This is really easy to see if you have root access to a fairly busy server, you’ll often see FIN packets coming from private IP address ranges, rather than the address of their NAT router, just run “tcpdump src net 192.168.0.0/16” to see this)
Something you might also notice is that windows servers are not affected by this, this is because windows machines close sockets very quickly if the computer at the other end becomes unresponsive, where as on a linux machine it takes anywhere between 13 and 30 minutes. (I’d be very interested to see if the same problem is as common on the apoc servers which run on windows).
- Kick people based on a timeout since the last time the server received data from the client. (This is exactly what we have currently I believe).
- Tweak the OS tcp parameters to close sockets more quickly should the remote end of the socket become unresponsive (The major downside to this is that it will affect all tcp connections to the server, rather than just scorched).
- Kick people when the server fails to send data to them. (This works, unless you try to send data faster than the client’s network connection)
I really don’t think it’s a problem with scorched3d, as you can alter the timeout values in your server.xml configuration file. (though very short timeouts sometimes lead to kicking people falsly if the server is running very slowly for some reason)
By default on linux if a client becomes unresponsive, it will try to send 15 more packets to the client, with each packet it will increase the amount of time to wait before sending the next one, according to the kernel documentation 15 retries gives a time between 13 and 30 minutes depending on the retransmission timeout (this is altered dynamically depending upon the latency of the client and other networking voodoo)
This can be altered via the sysctl sys.net.ipv4.tcp_retries2 (if you’re into that network voodoo stuff).
Finally got around to doing a little bit more on the spitfire (took so long due to a computer crash)
I’ve still got a few details to add on, but it’s coming along nicely (after I reinstalled blender, gimp, etc since my computer died :()
Ok, I’ve skinned a cube for ya.
It took me ages to pick up the UV editing in blender, editing models was bad enough.
It’s basically the same as vertex editing, except half the commands don’t work (e.g. you can’t control select) and you have to use the keyboard, no context menus 🙁
I was going to try to export it so scorched3d could use it, but I’ll have to get around to installing cbx’s plugin again to do this.
I notice quite a bit of lag at the start of a round where the server sends the map to all clients.
It should be possible (using a seeded pseudo random number generator for the code which generates the landscape) just to pass the seed, 4 or 8 bytes would quite easily suffice for this. (as opposed to the 256 * 256 * 4 bytes requires for an entire map)
It would also be really cool to be able to simulate errosion types to create different types of landscape.
It would be very cool to see things like river valleys, glacial features, cliffs, canyons in the generated landscapes.
Maybe I’ll get around to having a stab at this myself (bit busy in real life at the moment, but maybe in a few weeks time I might have some spare time).
Ooh, just remebered. It would be cool to fix the wind so it was more realistic, right now (I haven’t peeked at the code on this one), there is no air resistance (hence why with no wind trajectories are parabolic). I’m guessing that wind just works by altering the gravity (Otherwise with no air resistance, wind wouldn’t affect the flight path of projectiles).
You could even have different weaoons with different amounts of air resistance, which would allow you do really cool things like a scatter bomb which drops bomblets with parachutes (the bomblets would have high air resistance which would cause them to fall slowly).
The only downside with this would be that it would make aiming MUCH more difficult that it already is.
Another idea is that it would be good to be able to make weapons explode when they are falling and less than a certain alititude above the ground. (this would allow you to for instance make a funky type bomb which explodes in the air above where you aimed it).
The todo page is giving me an internal server error atm.
Speed control doesn’t work in multiplayer network games, only local games.
Not too much use for Director soupie, err Direwolf 🙁
Oh, and I’m back… My computer’s CPU fried
Somewhere (without going looking for it) there is a post from gavin saying they only update every 6 hours now.
If you want to see something really funny, move two tanks to the exact same position (it is possible), then.
Tank A fires a heavy riot bomb up at 1000 power.
Tank B fires a ton of dirt up at 1000 power.
(without parachutes turned on)
I’ve had really weird results doing this, tanks floating in mid air, kills & self kills with ton of dirt, one client showing a mountain with a tiny whole in the middle the other showing a mountain with a big hole.
First time I’ve textured something… so it’s taking a while 🙂
Plus I’ve been busy with other stuff, ty though.
I’ve more or less got the texture coords done, what’s a good program for making textures in ? I’m currently trying to use the gimp (which I don’t know so well, but then I don’t know blender that well either).