| 1 | OPy Compiler and Byterun
 | 
| 2 | ========================
 | 
| 3 | 
 | 
| 4 | The OPy compiler is a Python bytecode compiler written in Python.  See
 | 
| 5 | [Building Oil with the OPy Bytecode Compiler][oil-with-opy].  It's currently
 | 
| 6 | used to translate Python source code in Oil to `.pyc` files.
 | 
| 7 | 
 | 
| 8 | The `byterun/` directory is a fork of [byterun][].  It's an experiment for
 | 
| 9 | learning what it will take to write a minimal interpreter for Oil.  It can
 | 
| 10 | currently run all Oil unit tests, but isn't otherwise used.
 | 
| 11 | 
 | 
| 12 | [oil-with-opy]: http://www.oilshell.org/blog/2018/03/04.html
 | 
| 13 | 
 | 
| 14 | [byterun]: http://aosabook.org/en/500L/a-python-interpreter-written-in-python.html
 | 
| 15 | 
 | 
| 16 | ## 2022 Update: OPy Will be "Replaced" By mycpp / Pea
 | 
| 17 | 
 | 
| 18 | A bytecode interpreter isn't fast enough to run Oil.  We still have the
 | 
| 19 | double-interpretation problem.
 | 
| 20 | 
 | 
| 21 | ## Getting started
 | 
| 22 | 
 | 
| 23 | Do the "Quick Start" in "in https://github.com/oilshell/oil/wiki/Contributing .
 | 
| 24 | 
 | 
| 25 | Then build the `py27.grammar` file:
 | 
| 26 | 
 | 
| 27 |     $ make _build/opy/py27.grammar.pickle
 | 
| 28 | 
 | 
| 29 | After Oil is setup, we can try out OPy.  Run these commands (and let me know if
 | 
| 30 | any of them doesn't work):
 | 
| 31 | 
 | 
| 32 |     oil$ cd opy
 | 
| 33 |     opy$ ../bin/opyc run gold/hello_py2.py  # basic test of compiler and runtime
 | 
| 34 | 
 | 
| 35 | Compile Oil with the OPy compiler:
 | 
| 36 | 
 | 
| 37 |     $ ./build.sh oil-repo  # makes _tmp/repo-with-opy and _tmp/repo-with-cpython
 | 
| 38 | 
 | 
| 39 | Run Oil unit tests, compiled with OPy, under **CPython**:
 | 
| 40 | 
 | 
| 41 |     $ ./test.sh oil-unit
 | 
| 42 | 
 | 
| 43 | Run Oil unit tests, compiled with OPy, under **byterun**:
 | 
| 44 | 
 | 
| 45 |     $ ./test.sh oil-unit-byterun   # Run Oil unit tests, compiled with OPy, under CPython
 | 
| 46 | 
 | 
| 47 | Gold tests in `gold/` compare the output of CPython vs. byterun:
 | 
| 48 | 
 | 
| 49 |     $ ./test.sh gold
 | 
| 50 | 
 | 
| 51 | Oil spec tests under byterun (slow):
 | 
| 52 | 
 | 
| 53 |     opy$ ./test.sh spec smoke  # like $REPO_ROOT/test/spec.sh smoke
 | 
| 54 |     opy$ ./test.sh spec all    # like $REPO_ROOT/test/spec.sh all
 | 
| 55 | 
 | 
| 56 | FYI, they can be run manually like this:
 | 
| 57 | 
 | 
| 58 |     $ gold/regex_compile.py  # run with CPython
 | 
| 59 |     $ ../bin/opyc run gold/regex_compile.py
 | 
| 60 | 
 | 
| 61 | Demo of the speed difference between OSH under CPython and OSH under byterun:
 | 
| 62 | 
 | 
| 63 |     ./demo.sh osh-byterun-speed
 | 
| 64 | 
 | 
| 65 | ## OPy Compiler Regtest
 | 
| 66 | 
 | 
| 67 | This uses an old snapshot of the repo in `_regtest/`.
 | 
| 68 | 
 | 
| 69 |     ./regtest.sh compile
 | 
| 70 |     ./regtest.sh verify-golden
 | 
| 71 | 
 | 
| 72 | ## Notes on Three OPy Builds
 | 
| 73 | 
 | 
| 74 | - `$REPO_ROOT/_build/oil/bytecode-opy`: Bytecode for the release binary.  Built
 | 
| 75 |   by `Makefile`.
 | 
| 76 | - `$REPO_ROOT/opy/_tmp/repo-with-opy`: The entire repo with OPy.  For running
 | 
| 77 |   Oil unit/spec tests under byterun, etc.  Built by `./build.sh oil-repo`.
 | 
| 78 | - `$REPO_ROOT/opy/_tmp/regtest`: The snapshot of Python files in `opy/_regtest`
 | 
| 79 |   are compiled, so we are insensitive to repo changes.  Built by `./regtest.sh
 | 
| 80 |   compile`.
 | 
| 81 |   
 | 
| 82 | ## OPy Compiler Divergences from CPython
 | 
| 83 | 
 | 
| 84 | ### Lexer
 | 
| 85 | 
 | 
| 86 | - I don't remember where exactly, but I ran into a bug lexing the CPython test
 | 
| 87 |   suite.  IIRC, CPython's lexer was more lenient about adjacent tokens without
 | 
| 88 |   spaces than `tokenize.py`.
 | 
| 89 | - `heapq.py` had `-*- coding: latin-1 -*-`, which causes problems.  OPy
 | 
| 90 |   should require `utf-8` source anyway.
 | 
| 91 | 
 | 
| 92 | ### Parser
 | 
| 93 | 
 | 
| 94 | - I ran into a bug where a file like `d = {}`, without a trailing newline,
 | 
| 95 |   gives a parse error.  Adding the newline fixes it.
 | 
| 96 | - print statements aren't allowed; we force Python 3-style `print(x, y,
 | 
| 97 |   file=sys.stderr)`.  I think this is because the parser doesn't know about
 | 
| 98 |   `__future__` statements, so it can't change the parsing mode on the fly.
 | 
| 99 | 
 | 
| 100 | ### Bytecode Compiler
 | 
| 101 | 
 | 
| 102 | - I think there are no `LOAD_FAST` bytecodes generated?  TODO: Make a bytecode
 | 
| 103 |   histogram using `opy/misc/inspect_pyc`.
 | 
| 104 | - The OPy bytecode is bigger than the CPython bytecode!  Why is that?
 | 
| 105 | 
 |