This issue tracker has been migrated to GitHub ,
and is currently read-only.
For more information,
see the GitHub FAQs in the Python's Developer Guide.
Created on 2014年04月16日 18:59 by doko, last changed 2022年04月11日 14:58 by admin. This issue is now closed.
| Files | ||||
|---|---|---|---|---|
| File name | Uploaded | Description | Edit | |
| issue21264.diff | doko, 2014年05月06日 12:57 | |||
| Messages (8) | |||
|---|---|---|---|
| msg216553 - (view) | Author: Matthias Klose (doko) * (Python committer) | Date: 2014年04月16日 18:59 | |
the installation directory is non-writable, and the byte code files don't exist. [1/1] test_compileall test test_compileall failed -- Traceback (most recent call last): File "/usr/lib/python3.4/test/test_compileall.py", line 194, in test_no_args_respects_force_flag self.assertRunOK('-f', PYTHONPATH=self.directory) File "/usr/lib/python3.4/test/test_compileall.py", line 144, in assertRunOK *self._get_run_args(args), **env_vars) File "/usr/lib/python3.4/test/script_helper.py", line 69, in assert_python_ok return _assert_python(True, *args, **env_vars) File "/usr/lib/python3.4/test/script_helper.py", line 55, in _assert_python "stderr follows:\n%s" % (rc, err.decode('ascii', 'ignore'))) AssertionError: Process return code is 1, stderr follows: 1 test failed: test_compileall Re-running failed tests in verbose mode Re-running test 'test_compileall' in verbose mode ====================================================================== FAIL: test_no_args_respects_force_flag (test.test_compileall.CommandLineTests) ---------------------------------------------------------------------- Traceback (most recent call last): File "/usr/lib/python3.4/test/test_compileall.py", line 194, in test_no_args_respects_force_flag self.assertRunOK('-f', PYTHONPATH=self.directory) File "/usr/lib/python3.4/test/test_compileall.py", line 144, in assertRunOK *self._get_run_args(args), **env_vars) File "/usr/lib/python3.4/test/script_helper.py", line 69, in assert_python_ok return _assert_python(True, *args, **env_vars) File "/usr/lib/python3.4/test/script_helper.py", line 55, in _assert_python "stderr follows:\n%s" % (rc, err.decode('ascii', 'ignore'))) AssertionError: Process return code is 1, stderr follows: |
|||
| msg217979 - (view) | Author: Matthias Klose (doko) * (Python committer) | Date: 2014年05月06日 12:57 | |
what happens here: PYTHONPATH=$(pwd) python3.4 -X faulthandler -S -m compileall Skipping current directory Listing '/home/packages/python/3.4/x'... Compiling '/home/packages/python/3.4/x/foo.py'... Listing '/usr/lib/python3.4/'... Listing '/usr/lib/python3.4/plat-x86_64-linux-gnu'... Listing '/usr/lib/python3.4/lib-dynload'... so even when overwriting PYTHONPATH, the standard locations are searched and tried to compile, which fails when these are read only. Is this behaviour expected? The tests itself can be fixed by passing the the test directory as an argument, as done in other tests too. |
|||
| msg218085 - (view) | Author: Matthias Klose (doko) * (Python committer) | Date: 2014年05月07日 22:01 | |
adding unittest developers |
|||
| msg251031 - (view) | Author: Terry J. Reedy (terry.reedy) * (Python committer) | Date: 2015年09月18日 21:46 | |
Verified failure on installed 3.4.3, Win 7, with 3 rather than 1 error (all similar). test_no_args_compiles_path (test.test_compileall.CommandLineTests) ... FAIL test_no_args_respects_force_flag (test.test_compileall.CommandLineTests) ... FAIL test_no_args_respects_quiet_flag (test.test_compileall.CommandLineTests) ... FAIL test_pep3147_paths_doubleoptimize (test.test_compileall.CommandLineTests) ... ok First traceback ends with AssertionError: Process return code is 1, command line was: ['C:\\Programs\\Python34\\python.exe', '-X', 'faulthandler', '-S', '-m', 'compil eall'], stderr follows: followed by another ending with File "C:\Programs\Python34\lib\encodings\cp1252.py", line 19, in encode return codecs.charmap_encode(input,self.errors,encoding_table)[0] UnicodeEncodeError: 'charmap' codec can't encode character '\U00011111' in position 38: character maps to <undefined> Why is an astral char in the error message? Will try patch. |
|||
| msg251039 - (view) | Author: Terry J. Reedy (terry.reedy) * (Python committer) | Date: 2015年09月18日 23:04 | |
David, I am nosying you because this issue proposes to modify two test functions you added to test_compileall. See my prior message. -- For me, the test failed at every first occurrence of 'PYTHONPATH=' in the 3 functions. With one occurrence removed, the next in the same function failed. With all 6 removed, I get 1 error: FAIL: test_no_args_compiles_path (test.test_compileall.CommandLineTests) ---------------------------------------------------------------------- Traceback (most recent call last): File "C:\Programs\Python34\lib\test\test_compileall.py", line 179, in test_no_args_compiles_path self.assertNotCompiled(self.initfn) File "C:\Programs\Python34\lib\test\test_compileall.py", line 159, in assertNotCompiled self.assertFalse(os.path.exists(path)) AssertionError: True is not false The tests all pass in current uninstalled 3.4.3+ repository. If I remove all 6, I get the error above. Changing how the directory is passed apparently changes what get compiled. From the title of the test ad the comment "# Note that -l is implied for the no args case.", this is expected and is the point of the test. So deleting 'PYTHONPATH=' must be wrong. I believe we should detect that modifying the environment will not work before calling assertRunOK and skip the latter if it cannot work. The other two functions were added in #19532 by RDM. I think we need to know why they were written as they are before changing them. Perhaps they pass with 'PYTHONPATH=' removed because they are incomplete. |
|||
| msg251099 - (view) | Author: R. David Murray (r.david.murray) * (Python committer) | Date: 2015年09月19日 18:45 | |
Yes the point is to test compilation of the path. If I run python3 -m compilelall on my installed python, it works fine (zero return code). I can't really investigate this further right now, but it isn't obvious to me what the answer is from the information in the issue so far, except that it doesn't *look* like it should be necessary to skip the tests unless the installation directories are read only *and* the .pyc files in them are non-existent or out of date. |
|||
| msg251610 - (view) | Author: Brett Cannon (brett.cannon) * (Python committer) | Date: 2015年09月25日 20:53 | |
Issue #25099 seems to also be related to a similar issue. A proposed fix can be found over there. |
|||
| msg252670 - (view) | Author: Brett Cannon (brett.cannon) * (Python committer) | Date: 2015年10月09日 22:11 | |
I consider this fixed by way of issue #25099. If that's wrong then feel free to open the issue again. |
|||
| History | |||
|---|---|---|---|
| Date | User | Action | Args |
| 2022年04月11日 14:58:02 | admin | set | github: 65463 |
| 2015年10月09日 22:11:37 | brett.cannon | set | status: open -> closed versions: + Python 3.5, Python 3.6 superseder: test_compileall fails when run by unprivileged user on installed Python messages: + msg252670 resolution: duplicate stage: patch review -> resolved |
| 2015年09月25日 20:54:01 | brett.cannon | set | assignee: brett.cannon |
| 2015年09月25日 20:53:54 | brett.cannon | set | nosy:
+ brett.cannon messages: + msg251610 |
| 2015年09月19日 18:45:03 | r.david.murray | set | messages: + msg251099 |
| 2015年09月18日 23:04:41 | terry.reedy | set | nosy:
+ r.david.murray messages: + msg251039 |
| 2015年09月18日 21:46:39 | terry.reedy | set | type: behavior messages: + msg251031 nosy: + terry.reedy |
| 2014年05月07日 22:01:04 | doko | set | nosy:
+ ezio.melotti, michael.foord messages: + msg218085 |
| 2014年05月06日 12:57:28 | doko | set | files:
+ issue21264.diff messages: + msg217979 components: + Tests keywords: + patch stage: patch review |
| 2014年04月16日 19:00:12 | doko | link | issue17750 dependencies |
| 2014年04月16日 18:59:40 | doko | create | |