CEGUI (Crazy Eddie's GUI) Forums

Hello CEGUI community,

I am using Ogre 1.7 and the CEGUI SDK for MinGW. At first I succeeded in using CEGUI without problems but when I started to use a Pixmap font a segmentation fault occurs within the CEGUIImageSet.cpp in line 178. Since the SDK won't help much in debugging this issue I tried to compile CEGUI from its sources. But I got several problems here, too. (For compiling from sources I am using CEGUI's development branch and therefor CMake, because compiling 0.7-5 for Code::Blocks without CMake is a pain in the a**).
I am also compiling Ogre from its sources. I have one debug build and one release build. Both builds create a custom version of the OgreBuildSettings.h that don't lie in Ogre's source directory, so CEGUI can't find them when compiling its OgreRenderer. CEGUI also can't find boost's thread library that is needed by Ogre.

Is there anything I can do to fix that?


Greetings
BrightBit

If you can get a backtrace / callstack from the point of the segfault, that would be cool.

I have to state that using the development code may be a lot of hassle for you, though we are glad of any testing This is completely untested by us using mingw. The unstable is now using a new source based dependencies pack which you can get here: https://sourceforge.net/projects/crayze ... pendencies again, the build for these is untested on mingw.

For setting up the build for Ogre, I'm not 100% sure, since I've not done it on windows. As far as building from an Ogre 'build' directory, where all the headers are not in the same location, this will have to be solved by the user (by copying the file or whatever). With some luck Kulik can advise on how he goes about all this on Windows, since I believe he has done it there, though neither of us use Windows as our primary platform.

CE.

Hi CrazyEddie and thanks for your answer,

here's the backtrace:

CrazzyEddie wrote:neither of us use Windows as our primary platform.

I can't blame you for that.

Logs from the 0.8 run please The Imageset ptr is zero which indicates an invalid Image was created. The imageset referring format has changed so maybe it's related to that ("set:Imageset image:Image" is now "Imageset/Image").

Hi Kulik,

I didn't succeed in compiling 0.8 as I mentioned in my first post. But my first post contains the Log for 0.7.5 which is the version where I get the segfault.

Kulik wrote:The imageset referring format has changed so maybe it's related to that ("set:Imageset image:Image" is now "Imageset/Image").

Normal fonts (true type fonts) work with CEGUI. A Pixmap font is causing the segfault, so is there any difference between these two types of fonts regarding imageset creation?


Greetings
BrightBit

Would it be possible for you post the xml for the font Menu and the xml for the Imageset MenuFont. It appears that everything is loading ok, but the Imageset 'this' being 0 is worrying, since the Image that calls it is created by the imageset!

If you compile the samples in the SDK, does the FontDemo application work ok - especially selecting the FairChar font, since that's a pixmap based font?

CE.

CrazyEddie wrote:Would it be possible for you post the xml for the font Menu and the xml for the Imageset MenuFont.

Of course. Here they come.

Menu.font:


MenuFont.imageset:

CrazyEddie wrote:If you compile the samples in the SDK, does the FontDemo application work ok - especially selecting the FairChar font, since that's a pixmap based font?

I did test it now and it actually works. Hmmm... I will try to find the difference between the FairChar Sample and my version. Thank you for the hint. I didn't notice the Font sample before.

I was able to compile the sample and found something.

The main difference between my code and the code of the Font Sample is the method CEGUI::Window::setFont. The sample uses this method to explicitly set the font of a window. Whereas my code implicitly defines the font to use via *.scheme files. I changed my code to use the method setFont as well and it actually worked. Then in turn I changed the TaharezLook.scheme of the Font Sample to use the FairChar font instead of the DejaVuSans font and indeed the segfault I got in my app now also happened in the sample (or at least some similar error).

This is the backtrace of the sample when I got the segfault:



Greetings
BrightBit

I tested this by ensuring the font referenced in TaharezLook.scheme was the pixmap "FairChar-30" - the samples ran fine for me (aside from the missing glyphs, since that font only contains upper case chars ). certainly no segfault. This test was run on linux though - I no longer use Windows other than for playing some games.

CE.