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.

Author mattip
Recipients Bob, doko, larry, mark.dickinson, mattip, meador.inge, python-dev, rkuska, steve.dower
Date 2015-03-24.20:55:43
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1427230544.28.0.19062230518.issue20160@psf.upfronthosting.co.za>
In-reply-to
Content
This is the workflow for developing a patch for python
https://docs.python.org/devguide
This is the workflow for adding a patch to an issue
https://docs.python.org/devguide/patch.html

If there is an issue with python, please demonstrate it in a way that we can follow, and when you submit a patch please upload it in a way that we can see what version of the original file you are changing.

libffi is a library used in python to allow foreign function interfaces in python. It lives here https://github.com/atgreen/libffi The code from libffi was imported into python at version 3.1, (Modules/_ctypes/libffi) but since the official compiler used for windows is msvc, a forked version of libffi, called libffi_msvc lives in parallel to the official libffi in the python source tree. Clang should be using the libffi code or a libffi provided with clang, not the libffi_msvc version since it is specific for msvc. If the libffi inside python is faulty, someone should create an issue to import a newer version from the "upstream" source at https://github.com/atgreen/libffi

In addition, you reference an email from 2012, but now relate to an issue with python 2.7.9 which was released much after that date. What is the problem you see, how does it manifest itself? You should create a separate issue for that behaviour

This issue here was specifically for using libffi with the microsoft compiler to support ctypes, and was fixed for the 2.7.9 release. A test was added in the general ctypes test directory to make sure small and large structures are correctly handled in ctypes. While the test is in a file named win32, the test is run on all platforms, thus exposing a problem with aarch64 on redhat systems in Jan 2015. Their issue was that the libffi in python, and the one used on their platform, had an issue with aarch64. But the upstream libffi does support aarch64, so again, this issue is not the place to report that failure.

In summary, f there is a problem using clang with libffi, perhaps open a new issue that clearly shows how clang+libffi+python fail to do something, it may be enough to simply not use the msvc version of libffi
History
Date User Action Args
2015-03-24 20:55:44mattipsetrecipients: + mattip, doko, mark.dickinson, larry, meador.inge, python-dev, steve.dower, rkuska, Bob
2015-03-24 20:55:44mattipsetmessageid: <1427230544.28.0.19062230518.issue20160@psf.upfronthosting.co.za>
2015-03-24 20:55:44mattiplinkissue20160 messages
2015-03-24 20:55:43mattipcreate