Posts: 513
Threads: 10
Joined: May 2014
Thanks: 138
Given 89 thank(s) in 75 post(s)
Hey,
I wasn't able to compile the current source on debian, because STDERR_FILENO wasn't found. Had to include unistd.h in StackTrace.cpp for compiling.
Do i have to use a new version of gcc or any other ideas?
Ty,
Seadragon91
Posts: 6,485
Threads: 176
Joined: Jan 2012
Thanks: 131
Given 1074 thank(s) in 852 post(s)
@STR_Warrior: Actually the chunk in the middle is at height 255 for some reason; the usual chunk problem was the chunk being too low. Still no idea what happens there, I'll investigate once I fix the cSpawnPrepare segfault.
@ Seadragon91: It's possible that the include needs to be added, I just followed some online code and tried it out on my Ubuntu 12; other OSes may require additional stuff or even a different code path altogether. Feel free to
Posts: 952
Threads: 16
Joined: May 2013
Thanks: 66
Given 105 thank(s) in 89 post(s)
12-01-2014, 02:22 AM
(This post was last modified: 12-01-2014, 02:22 AM by tigerw.)
(11-30-2014, 10:54 PM)xoft Wrote: @Seadragon91: It's possible that the include needs to be added, I just followed some online code and tried it out on my Ubuntu 12; other OSes may require additional stuff or even a different code path altogether. Feel free to
...spontaneously combust.
Relevant username also.
Posts: 6,485
Threads: 176
Joined: Jan 2012
Thanks: 131
Given 1074 thank(s) in 852 post(s)
oh well, tabbed browsing, I start typing an answer, look sideways for some info, find another thing to write, needing another piece of info.... and when I get back through the chain, I post without reading
@ Howaner already fixed it: https://github.com/mc-server/MCServer/co...ada85fadff
Posts: 6,485
Threads: 176
Joined: Jan 2012
Thanks: 131
Given 1074 thank(s) in 852 post(s)
I wouldn't have believed it, but I'm debugging on Linux big time nowadays I found a nice piece of software, MobaXterm, that provides an SSH terminal together with X11 server for windows, so I can connect to my Linux laptop and start qtcreator over this connection and have it displayed on my desktop's big screen. Now, qtcreator is not as great as MSVC, but it's much better than editting individual files by an editor and launching make to compile. I won't even mention the ugly gdb interface; qtcreator is nowhere near MSVC here, but still is a thousand times better than command line
And I think I've finally found the reason for the "empty spawn chunk" problem using this weird setup.
Posts: 513
Threads: 10
Joined: May 2014
Thanks: 138
Given 89 thank(s) in 75 post(s)
12-01-2014, 04:29 PM
(This post was last modified: 12-01-2014, 11:55 PM by Seadragon91.)
Do you know far lands in mincecraft?
http://minecraft.gamepedia.com/Far_Lands
Very funny thing, this appears in older versions of minecraft.
I teleported me to this coords in mc server:
30000000 100 30000000
This thing has appeared in my world :
http://minecraft.gamepedia.com/World_border
I am not able to move anymore.
Edit: Okay did a few more tests in mc server. The generator do crazy things at 50k blocks
http://i.imgur.com/BagGwTY.png
Edit 2: Opened a own thread for that problem
Posts: 6,485
Threads: 176
Joined: Jan 2012
Thanks: 131
Given 1074 thank(s) in 852 post(s)
12-02-2014, 02:32 AM
(This post was last modified: 12-02-2014, 02:32 AM by xoft.)
I decided to give Lua Multithreading one more chance, so I tried to measure the performance impact of the mutex locking required for the thread safety, considering our current code. For this, I adapted code from my LuaMultiThreaded project ( https://github.com/madmaxoft/LuaMultiThreaded ) and let it measure the overall count of LuaLock calls. The numbers are bad:
Server startup and immediate shutdown: ~430'000 locks
One webadmin page requested: 1'500 - 2'500 locks
Initializing the Core: 5'000 locks
Initializing the Gallery: 24'000 locks
Core's OnTick handler (for the "tps" command"): 68 locks each call
I'm afraid that making this run multithreaded will kill the server performance-wise. The locks are just too much contended. Unfortunately Lua's locking is too fine-grained and thus doesn't scale well.
I have pushed my code to the LuaMultiThreadedLockCount branch; I don't expect it to be merged to master, in fact, I don't even expect it to compile under Linux.
Posts: 4,628
Threads: 115
Joined: Dec 2011
Thanks: 693
Given 494 thank(s) in 423 post(s)
So If we want multithreading for plugins we probably have to make multiple LuaStates for a plugin? That's a bummer
Posts: 6,485
Threads: 176
Joined: Jan 2012
Thanks: 131
Given 1074 thank(s) in 852 post(s)
I don't really know, I have no idea how much of a performance bottleneck those locks mean. So unless I try it, I won't know. Additionally, the branch doesn't compile under Linux, so we'd need a fix for that as well. I can't access Travis now for an unknown reason, so I don't even see the error message.
Posts: 6,485
Threads: 176
Joined: Jan 2012
Thanks: 131
Given 1074 thank(s) in 852 post(s)
After reading up a bit on the topic, it seems that such lock counts are not that bad, actually. We just might get away with it.
Ref.: http://preshing.com/20111118/locks-arent...ention-is/
|