bpo-11191: skip the distutils test 'test_search_cpp' when AIX and not gcc as compiler, etc. by aixtools · Pull Request #5206 · python/cpython (original) (raw)
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service andprivacy statement. We’ll occasionally send you account related emails.
Already on GitHub?Sign in to your account
Conversation5 Commits250 Checks0 Files changed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.Learn more about bidirectional Unicode characters
[ Show hidden characters]({{ revealButtonHref }})
skip the distutils test 'test_search_cpp' when AIX and not gcc as compiler
as xlc et al do not support -o (as output). 'cpp' output is ALWAYS to stdout
also add additional logic Lib/test/init.py for when
sysconfig.customize_compiler returns [] rather than None
add the additional argument '-C' for AIX xlc cpp processing so comments are
included in the cpp output rather than replaced by a single ' ' (space) char.
https://bugs.python.org/issue11191
A) This 'fixes' the distutils test 'test_search_cpp' by skipping it (on AIX) when the compiler is not gcc. When IBM XLC is used and the argument -E is applied, the argument -o is not accepted.
B) Even though the test is skipped, the processing of the 'pp_args' is corrected (-o is not added, -C is added so that comments are included in the output)
C) There was an additional error that caused 6 tests to fail (when sysconfig.customize_compiler returned [] rather than None). This is also corrected.
I am not claiming that this is the best solution. However, it does show where work needs to be done. Improvements are welcomed.
The 'wikipedia' example is now 'rosette', describing what it draws. The 'penrose' print output is reduced. The 'tree' '1024' output is eliminated.
Fix passing lists and tuples of strings containing special characters '"', '\', '{', '}' and '\n' as options to tkinter.ttk widgets.
flagsis indeed deprecated, but there is a validation on its value for backwards compatibility reasons. This adds mention of this in the docs.- The docs say that
sizehintis deprecated and ignored, but it is still used whenepoll_create1()is unavailable. This adds mention of this in the docs. sizehint=-1is acceptable again, and is replaced withFD_SETSIZE-1. This is needed to have a default value available at the Python level, sinceFD_SETSIZEis not exposed to Python. (see: bpo-31938)- Reject
sizehint=0since it is invalid to pass on toepoll_create().
The relevant tests have also been updated.
They can be exposed when some C API calls fail due to lack of memory.
- Failed Py_BuildValue() could cause an assertion error in the following TextIOWrapper.tell().
- input_chunk could be decrefed twice in TextIOWrapper.seek() after failed Py_BuildValue().
- initvalue could leak in StringIO.getstate() after failed PyDict_Copy().
Using OptiPNG 0.7.7.
Used command is: find . -name '*.png' | xargs optipng -o7
The line-length limit is not needed because the pages appear in a separate app rather than on a browser tab. It can also interact badly with the DPI setting.
When hunting memory leaks using -R 3:3, test_imap_unordered() of test_multiprocessing leaks randomly a few memory blocks. It is a false alarm: when testing using -R 3:20 for example, no leak is detected.
Modify test_imap_unordered() to be closer to test_imap():
- Only test 10 numbers instead of 1000: it's a pool of 4 processes, so 10 is enough to test at least one number per process
- Use chunksize=100 instead of chunksize=53 to mimick test_imap()
The Opera Browser was using a outdated command line invocation that resulted in an incorrect URL being opened in the browser when requested using the webbrowser module.
- Correct the arguments passed to the Opera Browser when opening a new URL.
PyErr_Print() will not return when the exception is a SystemExit, so decref the main module object in that case.
Without tp_clear, GC can't break cyclic reference. It will cause memory leak when cyclic reference is created intentionally.
PyErr_Print() always clears the error indicator, so there is no need to call PyErr_Clear() immediately afterwards.
posixmodule.c always declares environ, so don't bother catching a NameError in os.py.
Co-authored-by: Antti Haapala antti.haapala@anttipatterns.com
Get insync with master before making modifications
Closing the PR - as the merge with current master creates a lot of noise. Will resubmit in a cleaner form.
Nice. That will make it go away for anyone with the latest compiler - so now the patch would need to be even more specific - like many macos ones, where the version of the os is important. (For this the versio nof the compiler).
However, as a private person (hobbyist) it will never work for me. I am, it seems, many versions behind (or did they finally sync the Fortran and XLC version numbers) - and my pockets are not deep enough. So my bot would always continue to fail on this - and one fail - ALL fail, as far as the bot results are concerned.
Sigh. Mustard after the meal has been eaten.
p.s. PR 8709 (#8709) is the open PR for this issue.
Starting in V16.1, the Fortran and XLC version numbers are finally synched up, so V16.1's predecessor is V13.1.3. Which version are you on currently? If you have an active license for an older version, you should be able to upgrade to the latest version without charge.
Reviewers
