bpo-32381: Add _PyRun_AnyFileObject()#23723
Conversation
pymain_run_file() no longer encodes the filename: pass the filename as an object to the new _PyRun_AnyFileObject() function. Add new private functions: * _PyRun_AnyFileObject() * _PyRun_InteractiveLoopObject() * _Py_FdIsInteractive()
|
test_ttk_guionly failed on Ubuntu. I re-run the jobs. |
| } | ||
| FILE *fp = _Py_wfopen(filename, L"rb"); | ||
| if (fp == NULL) { | ||
| char *cfilename_buffer; |
There was a problem hiding this comment.
Why was this part removed? Because on Solaris (or any other platform where wchar_t values don't correspond to Unicode values) it no longer works. I found this when test_nonexisting_script (test.test_cmd_line_script.CmdLineTest) started failing with this:
...
File "/builds/jkulik/buildbot-sources/components/python/pythonmaster/Python-3.10.0-a82f63f/Lib/subprocess.py", line 1018, in _translate_newlines
data = data.decode(encoding, errors)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xe6 in position 202: invalid continuation byte
But maybe I am missing something and the problem is somewhere else, that's why I'm asking.
Oracle Solaris and wchar_t: https://docs.oracle.com/cd/E36784_01/html/E39536/gmwkm.html
There was a problem hiding this comment.
Oh. I knew that someone will complain about that :-) I removed the code since it was ugly code that I wrote years ago. The correct fix is to rely on Python sys.stderr, not use C stderr. I wrote PR #23778 for that.
There was a problem hiding this comment.
Thanks! I tried test_nonexisting_script with your new PR #23778 and it works again.
pymain_run_file() no longer encodes the filename: pass the filename as an object to the new _PyRun_AnyFileObject() function. Add new private functions: * _PyRun_AnyFileObject() * _PyRun_InteractiveLoopObject() * _Py_FdIsInteractive()
4 participants
pymain_run_file() no longer encodes the filename: pass the filename
as an object to the new _PyRun_AnyFileObject() function.
Add new private functions:
https://bugs.python.org/issue32381