[cpyrt] Spell out the CPython names behind the Python 2 aliases by guitargeek · Pull Request #128 · compiler-research/cppjit

guitargeek · 2026-10-01T10:57:39Z

cpyrt.h still carried the cpyrt_PyText_*, PyInt_*, cpyrt_PyCapsule_* and CPPJIT__*__ aliases that let the sources build against both Python 2 and 3. With Python >= 3.12 required they are plain renames of the PyUnicode_, PyLong_ and PyCapsule_* APIs and the Python 3 dunder names, so use those directly and drop the aliases, together with the unused cpyrt_PyText_GetSize, PyIntObject, PyBuffer_Type and dict_lookup_func.

cpyrt_PyText_AsStringAndSize stays: unlike PyUnicode_AsUTF8AndSize it also accepts bytes.

The preprocessed sources are unchanged except in two places. PyObjectDir27.inc tested PyType_Check(obj) || PyClass_Check(obj), which collapses to PyType_Check(obj). CPPJIT_INITIALIZE_STRING stringifies its argument without expanding it, so gDiv held "CPPJIT__div__" rather than "__truediv__"; it now holds "__truediv__". Nothing reads gDiv.

cpyrt.h still carried the cpyrt_PyText_*, PyInt_*, cpyrt_PyCapsule_* and CPPJIT__*__ aliases that let the sources build against both Python 2 and 3. With Python >= 3.12 required they are plain renames of the PyUnicode_*, PyLong_* and PyCapsule_* APIs and the Python 3 dunder names, so use those directly and drop the aliases, together with the unused cpyrt_PyText_GetSize, PyIntObject, PyBuffer_Type and dict_lookup_func. cpyrt_PyText_AsStringAndSize stays: unlike PyUnicode_AsUTF8AndSize it also accepts bytes. The preprocessed sources are unchanged except in two places. PyObjectDir27.inc tested PyType_Check(obj) || PyClass_Check(obj), which collapses to PyType_Check(obj). CPPJIT_INITIALIZE_STRING stringifies its argument without expanding it, so gDiv held "CPPJIT__div__" rather than "__truediv__"; it now holds "__truediv__". Nothing reads gDiv.

aaronj0

Great to see that we no longer need the cpyrt_* helpers which were quite annoying when writing CPython extension code. Can you take a look at the concern on PyInt_FromLong -> PyLong_FromLong call sites. Otherwise looks good.

aaronj0 · 2026-10-01T11:36:36Z

  // special case to be used with arrays: return a Python int instead of str
  // (following the same convention as module array.array)
-  return PyInt_FromLong((long)*((signed char*)address));
+  return PyLong_FromLong((long)*((signed char*)address));


I think this special case is supposed to retain the return as PyInt: https://docs.python.org/3/library/array.html

It was literally defined to be the same in cpyrt.h:
#define PyInt_FromLong PyLong_FromLong

So where is the concern exactly? Maybe you're hinting at something else that is not directly related to this change?

Oh wow, that looks wrong.. Then yes it's not directly related to your change. Was PyInt_FromLong only in Python2? It just looked like a functional change

Yes correct, with Python 3 all the PyInt_* functions got replaced by PyLong_* equivalents.

aaronj0 · 2026-10-01T11:36:43Z

  // special case to be used with arrays: return a Python int instead of str
  // (following the same convention as module array.array)
-  return PyInt_FromLong((long)*((unsigned char*)address));
+  return PyLong_FromLong((long)*((unsigned char*)address));


aaronj0 · 2026-10-01T11:40:51Z

-#define cpyrt_PyText_AsString PyUnicode_AsUTF8
-#define cpyrt_PyText_AsStringChecked PyUnicode_AsUTF8
-#define cpyrt_PyText_GetSize PyUnicode_GetSize
-#define cpyrt_PyText_GET_SIZE PyUnicode_GET_LENGTH


This is a good quality of life improvement :)

For human and machine 🙂

aaronj0

LGTM! I'd also annotate [NFC] in the commit message when merging

github-actions · 2026-10-01T12:48:45Z

CppJIT CI has detected a new failure on workflow Test running on main while building 4ba4a0d at job "macos-26-intel/llvm22/py3.12/c++20 / run".

Full details are available at: https://github.com/compiler-research/cppjit/actions/runs/36859698686
Here is the relevant piece of the build log for the reference

Step "Build and test" failure:
  Python2_ROOT_DIR: /Users/runner/hostedtoolcache/Python/3.12.10/x64
  Python3_ROOT_DIR: /Users/runner/hostedtoolcache/Python/3.12.10/x64
  LLVM_DIR: /Users/runner/work/cppjit/cppjit/install/lib/cmake/llvm
  Clang_DIR: /Users/runner/work/cppjit/cppjit/install/lib/cmake/clang
##[endgroup]
(node:15118) [DEP0040] DeprecationWarning: The `punycode` module is deprecated. Please use a userland alternative instead.
(Use `node --trace-deprecation ...` to show where the warning was created)
With the provided path, there will be 1 file uploaded
Artifact name is valid!
Root directory input is valid!
(node:15118) [DEP0169] DeprecationWarning: `url.parse()` behavior is not standardized and prone to errors that have security implications. Use the WHATWG URL API instead. CVEs are not issued for `url.parse()` vulnerabilities.
Beginning upload of artifact content to blob storage
Uploaded bytes 224
Finished uploading artifact content to blob storage!
SHA256 digest of uploaded artifact zip is 67356d31449782477d22b87cc119ec2dee3e15e77d23ce0ba4a716c300c54bc0
Finalizing artifact upload
Artifact result-macos-26-intel-llvm22-py3.12-cxx20.zip successfully finalized. Artifact ID 11161292468
Artifact result-macos-26-intel-llvm22-py3.12-cxx20 has been successfully uploaded! Final size is 224 bytes. Artifact ID is 11161292468
Artifact download URL: https://github.com/compiler-research/cppjit/actions/runs/36859698686/artifacts/11161292468
Post job cleanup.
Post job cleanup.
[command]/usr/local/bin/git version
git version 2.55.0
Copying '/Users/runner/.gitconfig' to '/Users/runner/work/_temp/daef6b3f-e602-4aba-b8c0-9f43a9351f73/.gitconfig'
Temporarily overriding HOME='/Users/runner/work/_temp/daef6b3f-e602-4aba-b8c0-9f43a9351f73' before making global git config changes
Adding repository directory to the temporary git global config as a safe directory
[command]/usr/local/bin/git config --global --add safe.directory /Users/runner/work/cppjit/cppjit
[command]/usr/local/bin/git config --local --name-only --get-regexp core\.sshCommand
[command]/usr/local/bin/git submodule foreach --recursive sh -c "git config --local --name-only --get-regexp 'core\.sshCommand' && git config --local --unset-all 'core.sshCommand' || :"
[command]/usr/local/bin/git config --local --name-only --get-regexp http\.https\:\/\/github\.com\/\.extraheader
http.https://github.com/.extraheader
[command]/usr/local/bin/git config --local --unset-all http.https://github.com/.extraheader
[command]/usr/local/bin/git submodule foreach --recursive sh -c "git config --local --name-only --get-regexp 'http\.https\:\/\/github\.com\/\.extraheader' && git config --local --unset-all 'http.https://github.com/.extraheader' || :"
[command]/usr/local/bin/git config --local --name-only --get-regexp ^includeIf\.gitdir:
[command]/usr/local/bin/git submodule foreach --recursive git config --local --show-origin --name-only --get-regexp remote.origin.url
Evaluate and set job outputs
Set output 'summary'
Cleaning up orphan processes
##[warning]Node.js 20 is deprecated. The following actions target Node.js 20 but are being forced to run on Node.js 24: actions/checkout@v4, actions/setup-python@v5, actions/upload-artifact@v4. For more information see: https://github.blog/changelog/2025-09-19-deprecation-of-node-20-on-github-actions-runners/

vgvassilev · 2026-10-01T16:23:59Z

Not arguing against this change. The advantage of the wrappers could be that we have a central place to check for errors of different CRuntime operations and have better diagnostics or ignore logic.

guitargeek · 2026-10-01T16:33:57Z

Sure, that's always the advantages of wrappers in general. But that's a whole different topic from having isolated aliases for a random tiny subset from the CPtthon API just because of Python 2 compatibility.

guitargeek force-pushed the python2_aliases branch from 8576567 to b92f90e Compare October 1, 2026 11:00

aaronj0 reviewed Oct 1, 2026

View reviewed changes

aaronj0 approved these changes Oct 1, 2026

View reviewed changes

guitargeek merged commit 4ba4a0d into compiler-research:main Oct 1, 2026
25 checks passed

guitargeek deleted the python2_aliases branch October 1, 2026 12:07

Navigation Menu

[cpyrt] Spell out the CPython names behind the Python 2 aliases - #128

[cpyrt] Spell out the CPython names behind the Python 2 aliases#128
guitargeek merged 1 commit into
compiler-research:mainfrom
guitargeek:python2_aliases

guitargeek commented Oct 1, 2026 •

edited

Loading

Uh oh!

aaronj0 left a comment

Uh oh!

aaronj0 Oct 1, 2026

Uh oh!

guitargeek Oct 1, 2026

Uh oh!

aaronj0 Oct 1, 2026

Uh oh!

guitargeek Oct 1, 2026

Uh oh!

aaronj0 Oct 1, 2026

Uh oh!

aaronj0 Oct 1, 2026

Uh oh!

guitargeek Oct 1, 2026

Uh oh!

aaronj0 left a comment

Uh oh!

Uh oh!

github-actions Bot commented Oct 1, 2026

Uh oh!

vgvassilev commented Oct 1, 2026

Uh oh!

guitargeek commented Oct 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

guitargeek commented Oct 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

aaronj0 left a comment

Choose a reason for hiding this comment

Uh oh!

aaronj0 Oct 1, 2026

Choose a reason for hiding this comment

Uh oh!

guitargeek Oct 1, 2026

Choose a reason for hiding this comment

Uh oh!

aaronj0 Oct 1, 2026

Choose a reason for hiding this comment

Uh oh!

guitargeek Oct 1, 2026

Choose a reason for hiding this comment

Uh oh!

aaronj0 Oct 1, 2026

Choose a reason for hiding this comment

Uh oh!

aaronj0 Oct 1, 2026

Choose a reason for hiding this comment

Uh oh!

guitargeek Oct 1, 2026

Choose a reason for hiding this comment

Uh oh!

aaronj0 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions Bot commented Oct 1, 2026

Uh oh!

vgvassilev commented Oct 1, 2026

Uh oh!

guitargeek commented Oct 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

guitargeek commented Oct 1, 2026 •

edited

Loading