On Sun, Sep 11, 2011 at 8:04 PM, Mike Nerone <[email protected]> wrote:
> Hello, all! First an extremely brief introduction since this is my first > email to this list: I'm a developer at Rackspace, and I've been there since > 2005, and I'm currently working with the Cloudkick team. which brings me to > libcloud. :) > Hi Mike, welcome to the libcloud community :) > > I'm working on support for OpenStack Nova 1.1. My intent had been to > preserve existing support for 1.0, inheriting as much behavior as possible > for the 1.1 driver. If other versions will be inheriting from the OpenStack 1.0 driver and re-use most of the functionality, I don't really see a problem with keeping it. If we decide to keep it, we also need to think about the API for getting the OpenStack driver instance. It would be nice if you could just pass in an API version (1.0, 1.1, ...) when instantiating the driver and the constructor would instantiate a and return a proper class for you, instead of you doing it manually (wooo, a factory! :)). So in short - I'm only against it if there are no OpenStack 1.0 deployments out there and if it adds a lot of complexity (most of the code from the 1.0 driver can't be re-used). > Discussion in IRC brought to light that pquerna had > instead envisioned dropping support for 1.0, and only supporting 1.1. I'm > concerned that since the driver supports third-party deployments (it > accepts > an auth_host parameter), it's possible that there may be existing private > deployments of which we're not aware. However, if everyone agrees that > there > are no such deployments or that we don't need to make them a concern, then > the OpenStack 1.0 support starts to look a lot like cruft, and I'd agree it > should be dropped. We need to have a conversation about the direction here. > I don't think there's much difference in the difficulty factor, so what > does > everyone think is the *right* thing to do? > > Thanks much! > > Mike Nerone >
