This example demonstrate how to have libs with their own examples and/or unit-tests which are compilable directly in their root folder or only the library as part of a bigger project. the idea is to have the libraries as subprojects included into the main-app (also revision-control wise).
you can compile this example in this main-folder (and get the main-app) or you can start tundra in extlibs/libA and get the library-example-app.
this example application structure looks like that:
app-root (folder) |- extlibs (folder) | |- lib1 (folder) | | |- subfolders of lib1 | | |- sources of lib1 | | | |- lib2 (folder) | |- sourcefiles for app
the main-app consists of sourcefiles and one extra folder called "extlibs" with subrepositories (buildsystem-wise as well as rvc-wise). these libraries can contain example- and/or testcode, too.
so in a perfect world the buildsystem is living inside lib1 which is used to build all examples and tests and another one in the application root that includes the build-rules from the library and uses these.
In the library folder there’s tundra.lua, library.lua and examples.lua. tundra.lua includes library and examples as unit-files. because when the library is compiled in the root-folder (examples from the library itself) or as a subfolder of "extlibs" the buildsystem needs to know the base-path in libraries.lua; here the option is used to set a variable in the global lua-table. in tundra.lua is written:
_G.LIBROOT_LIB1 = "."
and in the library.lua you can use this information as the base path:
Sources = {
_G.LIBROOT_LIB1 .. "/file.cpp" ,
Glob { Dir = _G.LIBROOT_LIB1 .. "/", Extensions = {".h"}},
...
now when you want to include the library as part of another application you can set the basepath in the variable in tundra.lua of the main-application accordingly:
_G.LIBROOT_LIB1 = "extlibs/lib1"
If there is the need to normalize the path this might help you (from John Leidegren):
local path = require "tundra.path" local LIBROOT_QBGFX = _G.LIBROOT_QBGFX -- or use native.getenv --I also normalize the path like this if LIBROOT_QBGFX ~= "." then LIBROOT_QBGFX = path.normalize(LIBROOT_QBGFX) end LIBROOT_QBGFX = LIBROOT_QBGFX .. "/" -- ensure end with trailing slash
Now if you have a base library (libBase) which is needed by Lib1 AND by the main application only adding it as a dependency to Lib1 is not enough, you also have to add it as dependency to the main application!
When using some sort of code generator with custom defrules which is provided by the library Lib1 and you want to be able to use this rule also in the mainapplication you have to add the rule also to the global lua table (Here we’re using some string-concatenation to build different defrules for a shader compiler, below is the complete block; notice the "function _G.compileVertexShader…"):
local function firstToUpper(str)
return (str:gsub("^%l", string.upper))
end
local function generateShaderCompilerDefRule(build, type, platform, path, profile, addParam)
local addOption = ""
if build == "debug" then
addOption = "--debug"
end
return DefRule {
Name = "ShaderCompiler"..firstToUpper(type)..firstToUpper(platform)..firstToUpper(path)..firstToUpper(build),
Pass = "CodeGeneration",
Command = "$(SHADERCOMPILER) "..addOption.." --type "..type.." --platform "..platform.." -p "..profile.." "..addParam.." -f $(<) -o $(@) -i ".._G.LIBROOT_QBGFX.."/src",
ImplicitInputs = { "$(SHADERCOMPILER)" },
Blueprint = {
Source = { Required = true, Type = "string", Help = "Input filename", },
Output = { Required = true, Type = "string", Help = "Output filename", },
},
Setup = function (env, data)
return {
InputFiles = { data.Source },
OutputFiles = { "$(OBJECTDIR)/shaders/" .. path .. "/" .. data.Output },
}
end,
}
end
generateShaderCompilerDefRule("debug", "fragment", "linux", "glsl", "120", "")
generateShaderCompilerDefRule("debug", "vertex", "linux", "glsl", "120", "")
generateShaderCompilerDefRule("debug", "fragment", "linux", "gles", "100", "")
generateShaderCompilerDefRule("debug", "vertex", "linux", "gles", "100", "")
generateShaderCompilerDefRule("release", "fragment", "linux", "glsl", "120", "")
generateShaderCompilerDefRule("release", "vertex", "linux", "glsl", "120", "")
generateShaderCompilerDefRule("release", "fragment", "linux", "gles", "100", "")
generateShaderCompilerDefRule("release", "vertex", "linux", "gles", "100", "")
function _G.compileVertexShader(inputFile, outputFile)
return {
{Config="*-*-debug"; ShaderCompilerVertexLinuxGlslDebug { Source = inputFile; Output = outputFile;}},
{Config="linux_imx6-*-debug"; ShaderCompilerVertexLinuxGlesDebug { Source = inputFile; Output = outputFile;}},
{Config="*-*-release"; ShaderCompilerVertexLinuxGlslRelease { Source = inputFile; Output = outputFile;}},
{Config="linux_imx6-*-release"; ShaderCompilerVertexLinuxGlesRelease { Source = inputFile; Output = outputFile;}},
}
end
function _G.compileFragmentShader(inputFile, outputFile)
return {
{Config="*-*-debug"; ShaderCompilerFragmentLinuxGlslDebug { Source = inputFile; Output = outputFile;}},
{Config="linux_imx6-*-debug"; ShaderCompilerFragmentLinuxGlesDebug { Source = inputFile; Output = outputFile;}},
{Config="*-*-release"; ShaderCompilerFragmentLinuxGlslRelease { Source = inputFile; Output = outputFile;}},
{Config="linux_imx6-*-release"; ShaderCompilerFragmentLinuxGlesRelease { Source = inputFile; Output = outputFile;}},
}
end
and now these generated defrules can be used(in units.lua from main-app):
Program {
Name = "test",
Sources = {
...
compileVertexShader("vs_cubes.sc", "vs_cubes.bin"),
compileFragmentShader("fs_cubes.sc", "fs_cubes.bin"),
},
Depends = {
...
There’s often the need to specify an include-directory to a library and also to the main application, this can be shortened this way:
Env = {
CPPPATH = {
"include",
"examples/common",
....
}
},
Propagate = {
Env = {
CPPPATH = {
"include",
"examples/common",
can be written as:
local repetetive_stuff = {
"include",
"examples/common",
....
}
...
Env = {
CPPPATH = {
repetetive_stuff,
}
},
Propagate = {
Env = {
CPPPATH = {
repetetive_stuff,
}
}
}