Trunk discussion / complains


Necromants also have +1 for troops of same alignment, so they got neutral morale on start.

PS It’s not a bug, I think. It’s a feature. Non-dead got no morale. And living troops of cause will get -1 to morale because of undead in army.


Ok, got it, thanks. :slight_smile:


I’ve manager to compile and run vcmi from git on my Linux, but the framerate seems to be extremely low. Is there any workaround?
vcmi from Ivan’s ppa work ok on the same machine.


Any details on hardware? Also can you clarify did you build it as debug or not?
Builds in PPA aren’t debug one as far as I know and debug may be quite slow.


4GM RAM, intel core i5-2410M (2.3GHz)
Ubuntu 14.10 x86_64
Intel integrated video. Regarding video: even Skyrim works with integrated video, so there is no possibility to complain about its performance. :slight_smile:
OpenGL core profile version string: 3.1 (Core Profile) Mesa 10.3.2

Not sure what it builds by default but gdb says debug symbols are loaded, so it either Debug or RelWithDebugInfo.

vcmi from ppa seems to be built at least as RelWithDebugInfo.

However, it seems like my own build uses software rendering or something similar. It feels like 1 frame per several seconds.


Personally I have same iGPU on laptop and HD4600 on desktop, but still noticed VCMI may become laggy with debug builds.

You will only get actual debug build if you set “-DCMAKE_BUILD_TYPE=Debug”. So it’s likely with debug symbols, but not real debug build.

You can easily see if OpenGL used or not by attaching APITrace.


@KroArtem @SXX As for now VCMI is not using graphics acceleration. The problem sholud be somewhere else.


Checked again, it was built with Debug, not RelWithDebInfo. I didn’t provide any options to cmake, though.
Will rebuild and check now.


With RelWithDebInfo and SDL set to 1.2 everything works as usual - pretty fast.


I’ve compiled vcmi under linux using CLion IDE (in order to debug vcmi a bit) and tried to launch vcmiserver first, but there is a problem whilst launching:

Does vcmiserver require something additional to launch?


IIRC you need to specify at least network port as command line parameter. Besides, standalone server is not something that VCMI supports right now - just start client and let it start server properly.


If I’m launching vcmiclient, it starts, but when a new game is started, it says:

That’s logical, because vcmiserver is not in /usr/local/bin, vcmi_BINARY_DIR is set (practically) to the same place as launcher and client.[/quote]


Btw, I’ve had to comment

#	find_package(Qt5Widgets REQUIRED)


#	find_package(Qt5Network REQUIRED)

because cmake didn’t manage to find these packages. The question is whether we had to write FindQt.cmake because I have libqt5network5 and libqt5widgets5 installed.


So the answer seems to be “no”, since nobody has the same problems. :slight_smile:


KroArtem, the problem may arise when person who haven’t installed libqt5network-dev (or whatever its name) wiil try to build VCMI. CMake won’t return any errors but build will fail due to missing headers.


I’d like to try to fully automate Ubuntu releases. To do this, I will create new branch on git - “master”. This branch will contain only release builds ( and probably dev builds to avoid huge delays between releases).

Once new version is ready for release, “develop” should be merged into “master” and this revision should be tagged.
@Warmonger - if you wish to avoid all that git maintance work, you can send me PM or notify me in any other way and release Windows builds as usual.


Hi, AVS, could you please tell me if building VCMI after this commit won’t be possible for WinXP users?


Seems like these files are related only to Code::Blocks IDE, so it won’t hurt you. :slight_smile: However, WinXP sounds way too old, though I liked this OS years ago.


I`ve already reverted back to winxp.


One question there. Should we make “BattleAI” default instead of “StupidAI”?