television/README.txt - llvm-archive - Git at Google

 LLVM-TV: LLVM Transformation Visualizer
 Written by: Misha Brukman, Tanya Brethour, and Brian Gaeke

 LLVM-TV is a tool that can be used to visualize the effects of transformations
 written in the LLVM framework. Compilation units in LLVM have a simple,
 hierarchical structure: a Module contains Functions, which contain BasicBlocks,
 which contain Instructions. At the lowest level of this hierarchy, Instructions
 may reference other BasicBlocks (with branches) or Functions (with calls),
 making the control-flow graphs and call graphs explicit. Our first task is to
 develop an interactive browser for these graphs and hierarchies.

 Our target audience is compiler developers working within the LLVM framework,
 who are trying to understand and debug transformations.

 Currently, the visualization tool will not reflect dynamic updates
 to the code; rather, it reflects the state of a compilation unit at
 a single instant in time, between transformations. Future work should
 extend this tool to reflect a dynamically updating view of the program as
 a transformation is modifying the code.

 How to compile:

 1. You must have wxWindows installed on your system, and wx-config has to be
    in your path.

    Make absolutely sure that wxwindows's configure picks up the same
    C++ compiler that you're using for llvm. Otherwise, you may get
    weird link errors when trying to link the llvm-tv tool.

 2. Configure and compile llvm-tv (you need an LLVM source and build trees):

 % cd path/to/llvm-tv
 % ./configure --with-llvmsrc=[path] --with-llvmobj=[path]
    If you're building in llvm/projects/llvm-tv, then you don't need
    to specify these --with options.
 % gmake

 Example of usage:

 % llvm-tv.exe &
    The .exe is not a typo; this command starts up the visualizer in
    the background using its wrapper script.
 % opt-snap -debug -licm -snapshot -gcse -snapshot < bytecode-file.bc > /dev/null
    This runs the llvm optimizer driver with the snapshot pass loaded, using
    another wrapper script, and makes two snapshots, which should appear
    in your visualizer.
	LLVM-TV: LLVM Transformation Visualizer
	Written by: Misha Brukman, Tanya Brethour, and Brian Gaeke

	LLVM-TV is a tool that can be used to visualize the effects of transformations
	written in the LLVM framework. Compilation units in LLVM have a simple,
	hierarchical structure: a Module contains Functions, which contain BasicBlocks,
	which contain Instructions. At the lowest level of this hierarchy, Instructions
	may reference other BasicBlocks (with branches) or Functions (with calls),
	making the control-flow graphs and call graphs explicit. Our first task is to
	develop an interactive browser for these graphs and hierarchies.

	Our target audience is compiler developers working within the LLVM framework,
	who are trying to understand and debug transformations.

	Currently, the visualization tool will not reflect dynamic updates
	to the code; rather, it reflects the state of a compilation unit at
	a single instant in time, between transformations. Future work should
	extend this tool to reflect a dynamically updating view of the program as
	a transformation is modifying the code.

	How to compile:

	1. You must have wxWindows installed on your system, and wx-config has to be
	in your path.

	Make absolutely sure that wxwindows's configure picks up the same
	C++ compiler that you're using for llvm. Otherwise, you may get
	weird link errors when trying to link the llvm-tv tool.

	2. Configure and compile llvm-tv (you need an LLVM source and build trees):

	% cd path/to/llvm-tv
	% ./configure --with-llvmsrc=[path] --with-llvmobj=[path]
	If you're building in llvm/projects/llvm-tv, then you don't need
	to specify these --with options.
	% gmake

	Example of usage:

	% llvm-tv.exe &
	The .exe is not a typo; this command starts up the visualizer in
	the background using its wrapper script.
	% opt-snap -debug -licm -snapshot -gcse -snapshot < bytecode-file.bc > /dev/null
	This runs the llvm optimizer driver with the snapshot pass loaded, using
	another wrapper script, and makes two snapshots, which should appear
	in your visualizer.