mirror of https://github.com/python/cpython
2c1e2583fd
A number of places in the code base (notably ceval.c and frameobject.c) rely on mapping variable names to indices in the frame "locals plus" array (AKA fast locals), and thus opargs. Currently the compiler indirectly encodes that information on the code object as the tuples co_varnames, co_cellvars, and co_freevars. At runtime the dependent code must calculate the proper mapping from those, which isn't ideal and impacts performance-sensitive sections. This is something we can easily address in the compiler instead. This change addresses the situation by replacing internal use of co_varnames, etc. with a single combined tuple of names in locals-plus order, along with a minimal array mapping each to its kind (local vs. cell vs. free). These two new PyCodeObject fields, co_fastlocalnames and co_fastllocalkinds, are not exposed to Python code for now, but co_varnames, etc. are still available with the same values as before (though computed lazily). Aside from the (mild) performance impact, there are a number of other benefits: * there's now a clear, direct relationship between locals-plus and variables * code that relies on the locals-plus-to-name mapping is simpler * marshaled code objects are smaller and serialize/de-serialize faster Also note that we can take this approach further by expanding the possible values in co_fastlocalkinds to include specific argument types (e.g. positional-only, kwargs). Doing so would allow further speed-ups in _PyEval_MakeFrameVector(), which is where args get unpacked into the locals-plus array. It would also allow us to shrink marshaled code objects even further. https://bugs.python.org/issue43693 |
||
|---|---|---|
| .. | ||
| buildbot | ||
| c-analyzer | ||
| ccbench | ||
| clinic | ||
| demo | ||
| freeze | ||
| gdb | ||
| i18n | ||
| importbench | ||
| iobench | ||
| msi | ||
| nuget | ||
| peg_generator | ||
| pynche | ||
| scripts | ||
| ssl | ||
| stringbench | ||
| test2to3 | ||
| tz | ||
| unicode | ||
| unittestgui | ||
| README | ||
README
This directory contains a number of Python programs that are useful
while building or extending Python.
buildbot Batchfiles for running on Windows buildbot workers.
ccbench A Python threads-based concurrency benchmark. (*)
demo Several Python programming demos.
freeze Create a stand-alone executable from a Python program.
gdb Python code to be run inside gdb, to make it easier to
debug Python itself (by David Malcolm).
i18n Tools for internationalization. pygettext.py
parses Python source code and generates .pot files,
and msgfmt.py generates a binary message catalog
from a catalog in text format.
iobench Benchmark for the new Python I/O system. (*)
msi Support for packaging Python as an MSI package on Windows.
parser Un-parsing tool to generate code from an AST.
peg_generator PEG-based parser generator (pegen) used for new parser.
pynche A Tkinter-based color editor.
scripts A number of useful single-file programs, e.g. tabnanny.py
by Tim Peters, which checks for inconsistent mixing of
tabs and spaces, and 2to3, which converts Python 2 code
to Python 3 code.
stringbench A suite of micro-benchmarks for various operations on
strings (both 8-bit and unicode). (*)
test2to3 A demonstration of how to use 2to3 transparently in setup.py.
unicode Tools for generating unicodedata and codecs from unicode.org
and other mapping files (by Fredrik Lundh, Marc-Andre Lemburg
and Martin von Loewis).
unittestgui A Tkinter based GUI test runner for unittest, with test
discovery.
(*) A generic benchmark suite is maintained separately at https://github.com/python/performance