1=========== 2Clang-Repl 3=========== 4 5**Clang-Repl** is an interactive C++ interpreter that allows for incremental 6compilation. It supports interactive programming for C++ in a 7read-evaluate-print-loop (REPL) style. It uses Clang as a library to compile the 8high level programming language into LLVM IR. Then the LLVM IR is executed by 9the LLVM just-in-time (JIT) infrastructure. 10 11Clang-Repl is suitable for exploratory programming and in places where time 12to insight is important. Clang-Repl is a project inspired by the work in 13`Cling <https://github.com/root-project/cling>`_, a LLVM-based C/C++ interpreter 14developed by the field of high energy physics and used by the scientific data 15analysis framework `ROOT <https://root.cern/>`_. Clang-Repl allows to move parts 16of Cling upstream, making them useful and available to a broader audience. 17 18 19 20Clang-Repl Usage 21================ 22 23 24.. code-block:: text 25 26 clang-repl> #include <iostream> 27 clang-repl> int f() { std::cout << "Hello Interpreted World!\n"; return 0; } 28 clang-repl> auto r = f(); 29 // Prints Hello Interpreted World! 30 31Note that the implementation is not complete and highly experimental. We do 32not yet support statements on the global scope, for example. 33 34 35Clang-Repl Basic Data Flow 36========================== 37 38.. image:: ClangRepl_design.png 39 :align: center 40 :alt: ClangRepl design 41 42Clang-Repl data flow can be divided into roughly 8 phases: 43 441. Clang-Repl controls the input infrastructure by an interactive prompt or by 45 an interface allowing the incremental processing of input. 46 472. Then it sends the input to the underlying incremental facilities in Clang 48 infrastructure. 49 503. Clang compiles the input into an AST representation. 51 524. When required the AST can be further transformed in order to attach specific 53 behavior. 54 555. The AST representation is then lowered to LLVM IR. 56 576. The LLVM IR is the input format for LLVM’s JIT compilation infrastructure. 58 The tool will instruct the JIT to run specified functions, translating them 59 into machine code targeting the underlying device architecture (eg. Intel 60 x86 or NVPTX). 61 627. The LLVM JIT lowers the LLVM IR to machine code. 63 648. The machine code is then executed. 65 66 67Just like Clang, Clang-Repl can be integrated in existing applications as a 68library (via using the clangInterpreter library). This turning your C++ compiler 69into a service which incrementally can consume and execute code. The 70**Compiler as A Service** (**CaaS**) concept helps supporting move advanced use 71cases such as template instantiations on demand and automatic language 72interoperability. It also helps static languages such as C/C++ become apt for 73data science. 74 75 76Related Reading 77=============== 78`Cling Transitions to LLVM's Clang-Repl <https://root.cern/blog/cling-in-llvm/>`_ 79 80`Moving (parts of) the Cling REPL in Clang <https://lists.llvm.org/pipermail/llvm-dev/2020-July/143257.html>`_ 81 82`GPU Accelerated Automatic Differentiation With Clad <https://arxiv.org/pdf/2203.06139.pdf>`_ 83
