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 vstinner
Recipients Vladimír Čunát, christian.heimes, heroxbd, vstinner
Date 2017-08-22.14:04:34
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1503410674.74.0.426730547414.issue31255@psf.upfronthosting.co.za>
In-reply-to
Content
If I understand correctly, Python 3.4 and 3.5 simply don't work on Linux < 3.17 with glibc >= 2.25.

The question is which OS make such change. It seems like the issue is more with binary packages when the builder is older or newer than the OS used by end users.

Since the bug prevents to start Python and cannot be worked again, maybe we need to fix Python 3.4... *but* this branch becomes old, Python 3.5 recently reached the security fixes only.

The pratical issue is that my changes os Python/random.c are large. I refactored the code to fix different issues.

I suggest to fix how binary packages are compiled, upgrade to Python 3.5 or newer, and more generally try to avoid the bad version combo (Linux < 3.17 with glibc >= 2.25) ;-)

I wrote articles on my os.urandom() changes:

* https://haypo.github.io/pep-524-os-urandom-blocking.html
* https://haypo.github.io/contrib-cpython-2017q1.html § os.urandom() and getrandom()
History
Date User Action Args
2017-08-22 14:04:34vstinnersetrecipients: + vstinner, christian.heimes, Vladimír Čunát, heroxbd
2017-08-22 14:04:34vstinnersetmessageid: <1503410674.74.0.426730547414.issue31255@psf.upfronthosting.co.za>
2017-08-22 14:04:34vstinnerlinkissue31255 messages
2017-08-22 14:04:34vstinnercreate