Spring RTS Engine

YokoZar wrote:Just put a dot next to the slowest player when the game is running at full slowdown, then let the players decide what to do. Having to guess from CPU usage isn't helpful.

That won't work. In the heat of the battle, do you expect anyone to have time checking for small dots? ROFL

FYI, the CPU usage has a different meaning for those who will be running the multithreaded version. With multithreading, any CPU usage less than 100% means that your system can keep up the speed, while the single threaded version can start to slow down the game at any CPU usage really (usually arou...

I don't know if the engine should really be doing that, and such a marker should be able to show how much faster it'll go without that person. If it's something like 2%, kicking isn't a good idea. Good point. But focusing on one player at a time raises another problem - suppose we have two equally ...

When a speed down occurs, logging it in the console could be useful for host to manage players. As you can't see the speed downs, after 30mn of intensive gaming, you realise "WTF the game runs at 0.35 !? Is a single player screwing the whole game ?". It is even worse when u discover that ...

* The code in GameServer that adjusts the gamespeed should be based on the median player instead of the slowest player. Useless. Having lots of fast clients doesn't make the slowest faster. The purpose is to get rid of the slowest players. If I understand correctly, players will drop if they cannot...

Since many games get ruined because of low gamespeed, I'm opening this thread to discuss the issue. Low gamespeed is mainly caused by: * Players that have slow machines. * Accidental adjustments of the gamespeed during gameplay that go unnoticed. * Intentional adjustments of the gamespeed made by pl...

Test cases are as follows. 1. At the top of luaV_gettable, insert code that calls the following after sim has been running for a minute. luaG_typeerror(L, t, "index"); If it worked correclty, the errors should be spammed on the info console, but it will crash. 2. In luaD_rawrunprotected be...

Zerver isn't able to share a lua enviroment between 2 threads, so it isn't a problem of any callins, it is a problem of rendering in lua in general. But still it is a problem of lua enviroment sharing and totaly independent of lua itself ... (so it is zerver's problem and not the prob of any lua de...

lurker wrote:zerver, which callins are in the rendering thread? Is there a list anywhere?

No list, but I believe they are all prefixed with "Draw". DrawWorldPreUnit for instance.

Yes, it would be good if rendering code for features and projectiles was moved out of Sim, but someone has to actually do this. I posted to keep new code from being added in the wrong place, but I don't know if anyone will be motivated to actually refactor this rendering code to another place anyti...

In conjunction with the 77b6 release, I'd like to at least have a link to the multithreading version posted/mentioned somewhere so that regular users can find it and try it. http://planetspring.free.fr/spring/exec ... _R7052.zip I've been running it for a while now and I consider it playable, some s...

Let me rephrase: "Tested only with BA".

CA is confirmed *not working* because it makes OpenGL calls from Sim thread, via LuaOpenGL I think. Multithreading Spring engine does currently not allow that.

For those who wanna test, try this one:

http://planetspring.free.fr/spring/exec ... _R7100.zip
http://planetspring.free.fr/spring/unit ... _R7100.zip

It should sync with 77b5 but is BA compatible only.

Edit: New version, should work with XTA also.

Aha, so the map is special. If you try the same trick with a vector you are dead meat.

I'm chasing this bug elsewhere then...

I've had some crashes pointing to explosiongenerator. It seems to store a pointer to a CEGData element in a static std::map, which is illegal to do if the data structure is going to be modified - and it is... cachedCEGs[tag] = cegData; currentCEG = &cachedCEGs[tag]; // yikes To me this bug seems...

I honestly thought the missing boxes were a bug until I saw this post...

Many ppl like to disable LuaUI to gain FPS. I know almost nothing about LUA, but the depth of the stacktraces I've stumbled across seem to indicate that it can eat CPU cycles.

I've seen that one too. U need some skills with Task Manager to get rid of the spam

Startboxes is so essential it should not be LUA based imo.

Yes, I'm assuming there will be a 77b6 soon since 77b5 seems to desync. I tried to play Tropical several times yesterday but couple of players would always instadesync.

With 77b6 we would have a good SMP testing opportunity.

imbaczek wrote:go to #buildserv, do !rebuild tag=spring_0.77b5 profile=gml

That profile does not use gmlsim=yes, so it will not have parallel Sim/Draw. Bibim, plz make a new profile...