* Parse JSON in Rust
* Reuse key when decoding JSON
* Unmark resolved test
* Parse null/true/false directly in call_scan_once
Parse JSON constants (null, true, false) directly in Rust within
call_scan_once() instead of falling back to Python scan_once.
This reduces Python-Rust boundary crossings for array/object values.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* Parse numbers directly in call_scan_once
Parse JSON numbers starting with digits (0-9) directly in Rust within
call_scan_once() by reusing the existing parse_number() method.
This reduces Python-Rust boundary crossings for array/object values.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* Parse NaN/Infinity/-Infinity in call_scan_once
Parse special JSON constants (NaN, Infinity, -Infinity) and negative
numbers directly in Rust within call_scan_once(). This handles:
- 'N' -> NaN via parse_constant callback
- 'I' -> Infinity via parse_constant callback
- '-' -> -Infinity or negative numbers via parse_constant/parse_number
This reduces Python-Rust boundary crossings for array/object values.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* Correct wrong index access
* Leave more flame span
* Refactor json scanstring with byte index
---------
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
* Use shared PYTHON_VERSION in cron-ci benchmark job
* Correct extra-tests/jsontests.py script
When the `extra_tests/jsontests.py` script was added, it was presumably during
the time when Python versions 3.5 to 3.8 were in use. At that time,
`test.libregrtest.runtest` was a valid path, but from CPython version 3.11
onwards, the path changed to `test.libregrtest.findtests`.
* Use ssl-rustls feature instead of ssl in cron-ci workflow
Replace the ssl feature with ssl-rustls in both the CARGO_ARGS
environment variable and the cargo-llvm-cov test command to fix
the cron-ci workflow.
Since 1a783fc9ec, it is disallowed
to use ssl manually.
* Replace inspect.getargspec with inspect.getfullargspec in custom_text_test_runner
inspect.getargspec() was removed in Python 3.11, causing jsontests.py to
fail with "TypeError: 'NoneType' object is not iterable" when running on
RustPython (which targets Python 3.13).
The get_function_args() function was silently catching the AttributeError
and returning None, which then caused the error in store_class_fields()
when trying to iterate over None.
> https://docs.python.org/3/whatsnew/3.11.html
> The getargspec() function, deprecated since Python 3.0; use
> inspect.signature() or inspect.getfullargspec() instead.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* Exclude rustpython-venvlauncher in cron-ci workflow
Since rustpython-venvlauncher is Windows-only,
it disables the project in the cron-ci workflow.
* Apply suggestion from @fanninpm
Co-authored-by: fanninpm <luxverdans@sbcglobal.net>
* Trigger cron-ci workflow in pull_request
---------
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: Jeong, YunWon <69878+youknowone@users.noreply.github.com>
Co-authored-by: fanninpm <luxverdans@sbcglobal.net>
* Update ruff to 0.14.11 in the CI
* Auto-format: cargo fmt --all
---------
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* Update test/libregrtest to CPython 3.13.10
* Re-add try block around platform and encodings info
* Disable WindowsLoadTracker on Windows
* Disable struct._clearcache()
* Disable saved_test_environment context manager
* Disable support.record_original_stdout()
* Update test/test_regrtest.py to CPython 3.13.11
* Add autotest shim from CPython 3.13.11
* Mark failing tests and skip flaky ones in test/test_regrtest.py
* Mark failing test in test_fstring
* Add _test_eintr.py from CPython 3.13.11
* Skip long-running test in _test_eintr
* Mark failing/erroring tests in _test_eintr
* Skip panicking tests in test_io
* Unmark passing tests in test_io
* Mark failing test in test_urllib2net
* Unmark passing test in test_fstring
* Time out if any one test takes longer than 10 minutes
* Skip more flaky tests in test_threading
* Skip tests that pollute the environment for some reason
* Skip flaky tests in test.test_multiprocessing_fork.test_misc
* Skip flaky tests in test.test_multiprocessing_fork.test_manager
Had to go with the nuclear option.
* Skip flaky tests in test_regrtest
* Mark flaky tests in test_ftplib
* Fix multiprocessing tests on Darwin
* Remove test_sys from list of environment polluters
* Remove test_ftplib from list of environment polluters
* Add test.test_multiprocessing_fork.test_misc to list of environment
polluters
* Check 10 times if a test is still polluting the environment
* Save more of the test environment before a test is run
* Mark test_import as polluting the environment
* Unmark passing test in test_regrtest
* Mark flaky test in test_regrtest
* Remove tests from list of environment polluters
* Skip flaky test in test_logging
* Remove tests from list of polluting tests
* Skip certain problematic tests in test_threading
* Skip test causing worker bug in test_builtin
* Make env polluter skip in test_threading os-agnostic
* Remove test_file_eintr from list of polluters
* Remove test_subprocess from the list of polluters
* Address coderabbit comments
* fix threading._dangling
* Skip environment polluter in test.test_concurrent_futures.test_thread_pool
* Make skip in test_thread_pool os-agnostic
* Skip problematic tests in test_wait
* Skip flaky tests in test_process_pool
* Remove test_import from list of polluters
* Amend skips in test.test_multiprocessing_forkserver.test_processes
* Remove test_raise from list of polluters
* Remove test_ssl from list of polluters
* Skip flaky test in test_ssl
* Fix YAML formatting issue
---------
Co-authored-by: Jeong, YunWon <jeong@youknowone.org>
