ciphercom

This would be a short fix (bit hacky) for the new Lion uint64 redefine error.

http://pastebin.com/meCvhM8z

Testing cmake-master on Ubuntu 64bit with gcc 4.5.1 using make (running "make -j2")

With included ACE and TBB:

[ 20%] Performing build step for 'ACE_Project'
make[3]: warning: -jN forced in submake: disabling jobserver mode.
cd /home/lynx/coding/mangos/dep/ACE_wrappers && /bin/bash /home/lynx/coding/mangos/dep/ACE_wrappers/aux_config/missing --run automake-1.11 --foreign
configure.ac:7284: required file `examples/Makefile.in' not found
configure.ac:7284: required file `examples/APG/Makefile.in' not found
(...)

Should also be fixed, but I don't have a gcc4.5 on my systems so I cannot check this. Like before only cmake-dev has this fixes. Also cmake-master will be deleted, as the cmake branch is exactly the same thanks to the backport of vmap3.

Testing cmake-master on Ubuntu 64bit with gcc 4.5.1 using make (running "make -j2")

With external ACE and TBB it doesn't link against tbb-malloc:

Linking CXX executable realmd
../framework/libframework.a(MemoryManagement.cpp.o): In function `operator new(unsigned long)':
MemoryManagement.cpp:(.text+0x5): undefined reference to `scalable_malloc'
(...)

With external ACE and included TBB, it doesn't link against libdl but needs to:

Linking CXX executable mangosd
/usr/bin/ld: ../game/libgame.a(ScriptMgr.cpp.o): undefined reference to symbol 'dlclose@@GLIBC_2.2.5'
/usr/bin/ld: note: 'dlclose@@GLIBC_2.2.5' is defined in DSO //lib64/libdl.so.2 so try adding it to the linker command line

Should be solved on branch cmake-dev. Will port it back to the stable cmake and cmake-master after all issues in this thread are cleared.

With this new build-system for the MaNGOS-Core projects master (/mangos/mangos) and one (/mangos/one) you can revolutionize the build-process of your server.

stable - https://github.com/cipherCOM/mangos/tree/cmake

nightly - https://github.com/cipherCOM/mangos/tree/cmake-dev

This project allows you to compile MaNGOS on your platform with your compiler and even your favorite IDE. This implies generation of project files for Visual Studio, Xcode, CodeBlocks, Eclipse CDT4 and KDevelop3 to name a few and therefore gain the full advantages for software development that each of these tools offer.

It all started with the cumbersome ./configure every time i wanted to set up my MaNGOS distribution. On the Unix-side this step is mandatory, really takes forever and the flood of options didn't make it any easier. I also disliked the idea to debug MaNGOS without any interface for GDB like you got one in Xcode (native IDE for Mac OSX). The windows-side clearly has the superior position when you want to do more than simply compile MaNGOS. But now, you even have the possibility on Unix to compile your server with PCH.

To give you an example on how I use CMake to generate a Xcode project for me, look at the following command:

mangosdir> mkdir build
mangosdir> cd build
build> cmake .. -G Xcode -DPREFIX=/opt/mangos-server -DPCH=1

Note: -G defines your generator, the type of IDE you want to use. Go and look for yourself if your IDE is supported http://www.cmake.org/cmake/help/cmake-2-8-docs.html#section_Generators

Of course, you can also use the GUI of CMake. So there is no need to open your terminal to generate the project files. And if I'm honest, I also use the GUI more often.

So give it a try, see if this CMake project can improve your daily working process with MaNGOS and enables you to try your favorite IDE.

But as with every young project everything is still in development. So I would appreciate if you can give me feedback about the system, new feature requests or errors you encounter so I can improve this system furthermore.

And one side note: The branch of this project also includes the needed sources for CMake to build the widely used scripting library and completely integrates it in the build process. So no more divided project files to compile every part of your server.

Thanks for your time and happy building... cC

And thanks to everyone that helped me so far... be it Funkybit, vladimir, DasBlub, Lynx3d, Zor, NoFantasy ... the list goes on!

Changed the computation of diff values in update cycles. Problem was when object were frozen and then updated again with not the 'real' diff.

Repository version was [s0368]

Author: me (github[at]ciphercom.de)

http://paste2.org/p/1067386

I calculated the 'real' diff time for Creature::Update(realDiff);. But now i would say the whole diff-computation system in mangos could be replaced with the timestamps... so a whole group of errors is fixed. Because the diff should always be calculated according to time i think.

First of all that fix only changes behavior in dungeons not the world. But Patman you are more or less right. The project to get 'groups' in mangos is already on my list but don't know when to implement it. This little patch fixes the behavior until then, because in dungeons all creatures should come.

For example: http://www.wowhead.com/spell=33822 but also others

Right, but the timer has to go down... or '5 minutes' respawn is not a correct way to specify respawntime.

TemporySummons didn't calculated their DespawnTimer correct when no player is around, because no Update(...) is called.

Repository version was [s0368]

Author: me (github[at]ciphercom.de)

http://paste2.org/p/1061309

Knockback of dx=0 and dy=0 (so no effect) and vmaps turned on crashes.

Repository version was [s0368]

Author: me (github[at]ciphercom.de)

http://paste2.org/p/1061303

Call for help in dungeons should call all hostile creatures and not only the same faction.

Repository version was [s0368]

Author: me (github[at]ciphercom.de)

http://paste2.org/p/1061301

Computation of SpellBonus with %-value was not correct. When two Spells both add %-Heal (or anything else) they computed the interest of interest (hope that is the right term). So two 10% heal add actually improved heal by 21% not 20%. With high stacking spells from some bosses this really added up.

Repository version was [s0368]

Author: me (github[at]ciphercom.de)

http://paste2.org/p/1061293

RespawnTime of killed NPC with loot was not correct when player left region of corpse. DecayTimer didn't get the Update(...) because no player was around. Same should be the case for the GroupLootTimer

Repository version was [s0368]

Author: me (github[at]ciphercom.de)

http://paste2.org/p/1061287

There are no const declarations in the InstanceSave class. So no const variables of this type are possible. Added const keywords to functions that don't change the object.

Repository version was [s0132]

Author: me (github[at]ciphercom.de)

http://paste2.org/p/966591

The speed difference is in my opinion too huge... Maybe you're don't cache the results and every time do a complete recompile? Or do you mean an complete recompile?

I could time my compile if it help's?... but I think it's about 10min for game and I have only an MacBook "13.