On Sun, Mar 20, 2016 at 4:39 PM, Volker Braun <[email protected]> wrote: > On Sunday, March 20, 2016 at 10:40:56 PM UTC+1, William wrote: >> >> That doesn't work since SMC uses normal Linux users, not lxc or docker >> containers, so they do not have a virtual chroot'd filesystem. > > > Well that explains why massive parallel compilation can grind everything to > a halt; Without cgroups its not possible to limit a user's total RSS, say.
I did *NOT* say that SMC does not using cgroups. The are several distinct concepts here: - cgroups - docker containers - lxc - chroot Please don't conflate them. > Still, you can have a process-private mount namespace even without cgroups > (using unshare). I didn't know how to do that. Thanks for letting me know. I'll look into whether unshare is robust enough yet. > >> >> For most development (on the library, not on packaging, which is the >> kind you probably do a lot of), a virtualenv (or python setup.py >> develop) local setup of the sage library, which overrides a >> system-wide install would be much better. > > > But then you still need to cythonize Sage and build the docs (wheels won't > work when depending on non-system libraries). Yes, but just building the Sage cython code only isn't too bad, is it? Also, when Sage is broken up into many smaller libraries, building the relevant Cython code will be less of a burden. Cythonizing isn't necessary -- but compiling of course is. (ccache might help?) William -- William (http://wstein.org) -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
