On 25 August 2017 at 17:35, Philippe Mathieu-Daudé <f4...@amsat.org> wrote: > Hi Stefan, > > On 08/25/2017 12:57 PM, Stefan Hajnoczi wrote: >> >> Many scripts can benefit from the standard library argparse module, which >> has >> improvements over the older optparse module. Unfortunately argparse was >> only >> shipped in Python 2.7 so we need a fallback for Python 2.6. > > > I probably missed some discussion about it, but what are the reasons to stay > 2.6 compatible? > > Python 2.6 support ended during October 2013, 4 years ago... [1] Why don't > kill it, start deprecating 2.7 which support will end in less than 3 years > from now [2], and move efforts to version 3...?
Our choices about our dependencies are generally driven by "what are the versions available on the oldest distros which we wish to support building QEMU on", which typically is whatever the long-term-support versions of Ubuntu, SUSE, Redhat, etc are. Has somebody checked what that means for our Python version requirements? (It would certainly be nicer if we could get away with bumping the version-req rather than including a big lump of code with yet-another-software-license, but maybe we can't.) > Apparently we expect a C compiler compatible with GCC >= 4.1 which was > released on 2006, before Python 2.5 :S > Then QEMU_BUILD_BUG_ON() try to use C11 feature... That's protected by a configure check for exactly this reason. thanks -- PMM
