Can we get a current release tag? Creating packages for in-the-middle commits is not good. If you don't feel like creating a real release, just make it a prerelease. The last one is 2 years old.

Forum thread: http://forums.librocket.com/viewtopic.php?f=2&t=874

The only thing that I changed in the base library is exposing Rocket::Core::ElementStyle as dllimport/export. That isn't completely necessary, but it is convenient; I can undo that if you want.

What I don't have is CMake scripts.

Precalculating kerning for all character pairs consumed a lot of RAM (O(N*N) complexity). For CJK fonts that contain tens thousands of glyphs it was a highly unpractical approach, resulting libRocket consuming >1.5GB of RAM. Instead, encapsulate FT_Get_Kerning call in FontFaceHandle::GetKerning and let FreeType handle kerning on the fly. In my measurements the performance difference is negligible, especially in conjunction with compiled geometry.

Hey,

I'm trying to add decorator at runtime for new dynamic elements. And seems that decorators definitions in style or via properties is just ignored.

Element* newItem = document->CreateElement("div"); newItem->SetProperty("background-image", "invader.png"); newItem->SetProperty("background-decorator","image");

The main developer of the game engine I'm working on accidentally removed PPI-based RCSS unit support. I need them in the engine now, so I'm adding them back here.

This request contains the ability to have different font providers. Rocket core does not depend from truetype anymore and a RocketTrueTypeLib has been created. A BMFont provider ( font create with AngelCode's BMFont tool ) is also provided.

This implementation does not support effects for the moment.

I try to run the luainvader sample (VS2012, Win7). I can compile everything with lua 5.2.3, but I cannot run the sample. There is an error in line 101:

Rocket::Controls::Lua::RegisterTypes(Rocket::Core::Lua::Interpreter::GetLuaState());

All these functions failed:

LuaType<ElementForm>::Register(L);
LuaType<ElementFormControl>::Register(L);
//Inherits from ElementFormControl
LuaType<ElementFormControlSelect>::Register(L);
LuaType<ElementFormControlDataSelect>::Register(L);
LuaType<ElementFormControlInput>::Register(L);
LuaType<ElementFormControlTextArea>::Register(L);
LuaType<DataFormatter>::Register(L);
LuaType<DataSource>::Register(L);
LuaType<ElementDataGrid>::Register(L);
LuaType<ElementDataGridRow>::Register(L);
LuaType<ElementTabSet>::Register(L);
//proxy tables
LuaType<SelectOptionsProxy>::Register(L);

Call stack:

RocketCoreLua.dll!luaH_newkey(lua_State * L, Table * t, const lua_TValue * key) Zeile 436   C
RocketCoreLua.dll!luaV_settable(lua_State * L, const lua_TValue * t, lua_TValue * key, lua_TValue * val) Zeile 152  C
RocketCoreLua.dll!lua_setfield(lua_State * L, int idx, const char * k) Zeile 764    C
> RocketCoreLua.dll!Rocket::Core::Lua::ExtraInit<Rocket::Core::Element>(lua_State * L, int metatable_index) Zeile 50    C++
RocketControlsLua.dll!Rocket::Core::Lua::ExtraInit<Rocket::Controls::ElementForm>(lua_State * L, int metatable_index) Zeile 80  C++
RocketControlsLua.dll!Rocket::Core::Lua::LuaType<Rocket::Controls::ElementForm>::Register(lua_State * L) Zeile 69   C++
RocketControlsLua.dll!Rocket::Controls::Lua::RegisterTypes(lua_State * L) Zeile 63  C++

...s C++11.

• Formatting - change design to favour fixing thread and exception safety over performance.
• Increase performance in some situations by the provision of move constructors and move assignment functions.
• Attempt to reduce string memory reallocations for replacement and concatenation by avoiding temporaries.
• Clear hash value where required such as in Erase.
• Don't clear hash value if not required for Replace.
• Null terminate in clear as if one does Assign("X") then Clear then CString() one expects null terminated string.
• Don't use Clear in StringBase itself to favour concatenation performance, only release memory on destruction.
• Hopefully make tolower/touppwer simpler/faster and reduce memory reallocation potential and preserve hash more often.
• Add exception to replace assert for out of memory condition. Both could be used if assert is debug only.
• Some small changes to reduce the need to cast to T* so much..
• Use T() in preference to '\0' in an attempt to get proper character type for wide char types etc..
• c++11 required. I use Windows but did try g++ and cmake with -DCMAKE_CXX_FLAGS=-std=c++11 which worked.

Sorry for single commit.

I'm aware that 1.3.0 is tagged as a "pseudo" release, but it would be nice if the dev and runtime binaries were packaged for use with that release (or a more recent one if it's coming soon). Right now, when a project takes a dependency on libRocket, the project developers need to build libRocket from source on their respective platforms. For someone new to building libRocket, the process looks something like this on Windows 8 with Visual Studio 12:

1. Download libRocket source (or clone using GitHub app).
2. In the shell of your choice, find the CMakeLists.txt (it's in Build) and run cmake.
3. Oh, looks like it has a Freetype dependency. Let's go to http://www.freetype.org/download.html and find the binaries we need.
4. Alright, binaries unpacked and put in my Visual Studio "VC" folder.
5. Oh, CMake still can't find it. Let's read FindFreetype.cmake and see what happened.
6. Hm, it doesn't include VC/lib/freetype2 and VC/include/freetype2. Let's run CMake Config to set those paths, or set a FREETYPE_DIR environment variable.
7. Alright! Now CMake works and produces libRocket.sln for VS2013. Hit "Build Solution" and away we go!
8. ...or not. Why do I have so many build errors? What does "error LNK2026: module unsafe for SAFESEH image" mean and why does it appear 10 times in a row?
9. According to StackOverflow (here: http://stackoverflow.com/questions/10599940/module-unsafe-for-safeseh-image-c), the Windows binary that I just downloaded for Freetype2 is out of date. Lo and behold, it's 2.4.2, while the latest version is actually 2.5.3. Guess now I need to build Freetype from source too...

...and so on.

Other libraries I depend on simply require that I download lib and include files and extract them someplace that CMake can find them. By comparison, I end up spending an order of magnitude longer getting libRocket set up on my machines (OS X and Windows 8). There are no binaries, and I can't actually find much documentation on the process (in fact, the only documentation assumes that you've already got binaries: http://librocket.com/wiki/documentation/C%2B%2BManual/GettingStarted). I'm patient enough to do this for my own project, but it's hard to convince other developers to volunteer on my project if the process looks like the one above just for one of my project's dependencies.

To demonstrate that this isn't a Windows-specific problem: Issue #90 produced a cryptic error that could have been avoided by knowing how to properly build Freetype on OS X, or by having the binaries available as a download on GitHub.

I appreciate that packaging these binaries for a series of platforms (and, in the process, determining which platforms do and don't get binary packages) isn't easy or quick. If supplying binaries is not feasible, then a list of steps would still save a lot of time, as new developers on libRocket-dependent projects could just read those instructions instead of learning the hard way how to get the binaries they need on each of their environments.

I encountered a problem with display: block and position: absolute and figured out that libRocket doesn't handle it at all (HTML + CSS code in browser shows one result and analogous code in RML + RCSS shows completely different one). So I tried to fix it (reading CSS2 Specification section 10.3.7). Hope these patches will be useful.

I've managed to get the Rocket Python bindings to compile on OS X but now _rocketcontrols.so will not import.

The modules are imported like so:

boost::python::import("_rocketcore");
boost::python::import("_rocketcontrols");

Only if the second import is present does the code throw an exception with the message RuntimeError: extension class wrapper for base class Rocket::Core::Element has not been created yet

The same thing happens in a Python interpreter:

$ python
Python 2.7.5 (default, Mar  9 2014, 22:15:05) 
[GCC 4.2.1 Compatible Apple LLVM 5.0 (clang-500.0.68)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import _rocketcore
>>> import _rocketcontrols
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
RuntimeError: extension class wrapper for base class Rocket::Core::Element has not been created yet

Does anyone have any idea about how I can debug this?

From an MSYS2 shell:

robotogre@DESKTOP MINGW64 ~/libRocket/Build 
$ cmake -G "MSYS Makefiles" -D CMAKE_MAKE_PROGRAM="/usr/bin/cmake.exe" -D FREETYPE_INCLUDE_DIRS="..\Dependencies\freetype\include;..\Dependencies\freetype\include\freetype2" -D FREETYPE_LIBRARY="..\Dependencies\freetype\lib\libfreetype.dll.a" .

Error:

CMake Error at C:/Users/robotogre/dev/msys64/mingw64/share/cmake/Modules/CMakeTestCCompiler.cmake:69 (message):
  The C compiler

    C:/Users/robotogre/dev/msys64/mingw64/bin/cc.exe

  is not able to compile a simple test program.

  It fails with the following output:

    Change Dir: C:/Users/robotogre/dev/msys64/home/tomsh/libRocket/Build/CMakeFiles/CMakeTmp

    Run Build Command(s):C:/Users/robotogre/dev/msys64/usr/bin/cmake.exe -f Makefile cmTC_95166/fast && CMake Warning:
      Ignoring extra path from command line:

       
       cmTC_95166/fast


    CMake Error: Unknown argument -f
    CMake Error: Run 'cmake --help' for all supported options.





  CMake will not be able to correctly generate this project.
Call Stack (most recent call first):
  CMakeLists.txt:52 (project)

The issue appears to be that the cmake '-f' argument has been deprecated:

remove [-f] <file>...
            Deprecated since version 3.17.

            Remove the file(s). The planned behaviour was that  if  any  of  the  listed  files
            already  do  not exist, the command returns a non-zero exit code, but no message is
            logged. The -f option changes the  behavior  to  return  a  zero  exit  code  (i.e.
            success)  in  such situations instead.  remove does not follow symlinks. That means
            it remove only symlinks and not files it point to.

            The implementation was buggy and always returned 0.  It  cannot  be  fixed  without
            breaking backwards compatibility. Use rm instead.

https://github.com/netpipe/RocketShip

may start my own fork of this to fix the scrollbar issue and hopefully start development back up again. what a useful API!

The error says missing DoAllocConsole(); at line 70 in main.cpp Also, there is a semicolon missing at line 122 in the file libRocket\Samples\shell\src\win32\ShellRenderInterfaceExtensionsOpenGL_Win32.cpp

I have tried modifying the font support to just create textures as required. They shouldn't all be pre-created.

You save memory this way and only create what is required.

There should be no limitations and I wish to use any UCS2 character...

Is anyone else working on removing the font-charset workaround?

So far I managed to make it build textures as required but it has caused weird issues such as the text size not being applied or the styles.

Please do not close this saying its not a workaround as any Unicode character should work without specifying a list of required characters...

Might need a separate post but for a feature emoji should be on the list I reckon!