I asked if we should drop support for Linux kernel 2.6. I now consider that no, we should not. It's not worth it.
A colleague proposed to setup a RHEL 6 buildbot which would test Python on Linux 2.6. 2018-01-19 10:26 GMT+01:00 Antoine Pitrou <solip...@pitrou.net>: > What is the problem with supporting Linux 2.6? It increases the code size base. This compatibility code has to be maintained. It would even be better to make sure that it's tested ;-) > Do we need to rely on newer features? (which ones?) My pull request which removed support for FreeBSD 9 and older was quite large and so it was interesting to do it: https://github.com/python/cpython/commit/13ff24582c99dfb439b1af7295b401415e7eb05b According to reactions on this thread, I'm not sure anymore that removing a few lines of C code is worth it compared to loosing support for Linux 2.6 which seems to be important for many users. Python has some fallback code for "recent" Linux features like SOCK_CLOEXEC, accept4(), getrandom(), epoll_create1(), open() and O_CLOEXEC, etc. The worst part is that Python has to check once per process that open() doesn't ignore O_CLOEXEC flag. It requires one extra syscall. But well, compared to the total number of syscalls just for "python3 -c pass", this syscall is likely negligible :-) Victor _______________________________________________ Python-Dev mailing list Python-Dev@python.org https://mail.python.org/mailman/listinfo/python-dev Unsubscribe: https://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com
