On Wed, May 25, 2016 at 12:47 PM, Tim Bell <tim.b...@cern.ch> wrote: > Slight concern on how to deploy on a RHEL system base as software > collections are non-trivial. >
But also worth keeping in mind, Craton and its dependencies (most significantly, MySQL or PostgreSQL, + Taskflow dependencies, including ZooKeeper for operators at scale) will only be deployed on nodes that are actually running Craton specifically, not at all on the fleet itself. > > > If we can keep the client to be still python 2.X compatible, that would be > a significant help. > Very good point about differentiating the (forthcoming) Python client from the rest of the Craton codebase. Like other projects, and as would be expected, the Python client (python-craton) is just going to be a wrapper of Craton's REST API. We will write it such that it is compatible for Python 2.7 and 3.4+; and tested accordingly. I guess we were thinking this implicitly - but now stated explicitly. > > > Getting good development productivity/deployments should probably outweigh > these concerns though… > I appreciate that. Python 2.7 is technical debt at this point; sometimes we have to support it if it makes sense (the client), but given that we have to support Python 3 as well, moving forward on that with a new project means less work/more productivity. - JIm
_______________________________________________ OpenStack-operators mailing list OpenStack-operators@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators