llvm-general-pure- Pure Haskell LLVM functionality (no FFI).

Safe HaskellSafe



Names as used in LLVM IR



data Name Source

Objects of various sorts in LLVM IR are identified by address in the LLVM C++ API, and may be given a string name. When printed to (resp. read from) human-readable LLVM assembly, objects without string names are numbered sequentially (resp. must be numbered sequentially). String names may be quoted, and are quoted when printed if they would otherwise be misread - e.g. when containing special characters.


means the seventh unnamed object, while


means the object named with the string "7".

This libraries handling of UnNames during translation of the AST down into C++ IR is somewhat more forgiving than the LLVM assembly parser: it does not require that unnamed values be numbered sequentially; however, the numbers of UnNames passed into C++ cannot be preserved in the C++ objects. If the C++ IR is printed as assembly or translated into a Haskell AST, unnamed nodes will be renumbered sequentially. Thus unnamed node numbers should be thought of as having any scope limited to the Module in which they are used.


Name String

a string name

UnName Word

a number for a nameless thing