On Tue, Sep 12, 2017 at 11:29:22AM -0700, Joshua Harlow wrote:
> Hi folks,
> 
> I know there is a bunch of usage of subprocess in openstack and especially
> since there is heavy usage of python 2.7 it made me wonder if we should try
> to move to subprocess32 to avoid some of the bugs that seem to exist (maybe
> distributors backported them?):
> 
> For example a major one (seems to be):
> 
> - https://github.com/google/python-subprocess32/commit/6ef1fea55
> 
> """Popen.wait() is now thread safe so that multiple
> 
>   threads may be calling wait() or poll() on a Popen instance at the same
> time
>   without losing the Popen.returncode value.
> """
> 
> That one concerns me slightly, because I know that certain openstack
> projects do use threads (and not eventlet monkey-patched green-thread
> hybrids).
> 
> TLDR; should we (could we?) switch?

We could.  It wouldn't be hard to propose a change to the requirements
repo and that could be used to test what breaks.

As to should we I'm not convinced.  It does give us a slightly more
modern subprocess module but it hasn't been updated in nearly 2 years.
I get that it's a backport from 3.3 which isn't getting updated but ...

Also it means adding something like:
if os.name == 'posix' and sys.version_info[0] < 3:
    import subprocess32 as subprocess
else:
    import subprocess

All over the place which isn't so great.

So overall I'm not certain it's worth it.

Yours Tony.

Attachment: signature.asc
Description: PGP signature

__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Reply via email to