[Openstack] Fwd: [nova] [rfc] drop XML from v3 API entirely
Jay Pipes
jaypipes at gmail.com
Thu Jan 16 15:20:35 UTC 2014
On Thu, 2014-01-16 at 07:22 -0500, Sean Dague wrote:
> On 01/15/2014 10:41 PM, Tom Fifield wrote:
> > On 15/01/14 22:02, Russell Bryant wrote:
> >> On 01/13/2014 08:52 PM, Tom Fifield wrote:
> >>> According to the most recent user survey, more than 30% of users consume
> >>> the XML API.
> >>
> >> Do you have a reference with more detail on this? Simply "30%" doesn't
> >> really mean much without more detail on what the question was, what the
> >> choices where, how many answered the question, etc ...
> >
> > Survey questions can of course be found at
> > https://www.openstack.org/user-survey/. This particular one was found on
> > the second page of the "About your OpenStack Deployments" section:
> >
> > """
> > What formats do you use for the API?
> > [X] XML
> > [X] JSON
> > """
> >
> > From the data from 2013-10-11:
> >
> > 30.67% of survey takers ticked XML
> > 69.32% of survey takers ticked JSON
> >
> > of course this question allowed overlap - the ticking of both boxes. The
> > overlap is 17.17% (or, if you prefer - 13.49% "pure" XML, and 52% "pure"
> > JSON)
> >
> >
> > 83% of deployments answered this question. As a comparison, this is a
> > similar number to how many answered the "hypervisors" question (87%).
>
> That also assumes that survey responders understood the question the way
> we are interpreting it. The actual question should have been:
>
> What toolkits do you use to interface with the OpenStack API?
> * OpenStack command line (i.e. nova cli)
> * openstack python libraries
> * jclouds
> * deltacloud
> ...
> * other - please specify
> * none - wrote our own
>
> If you wrote your own toolkit for accessing the OpenStack API what wire
> format are you using?
> * JSON
> * XML
>
> If you wrote your own toolkit for accessing the OpenStack API what
> programming language did you write it in?
> * C/C++
> * C#
> * Java
> * Python
> * Perl
> * PHP
> * Ruby
> * Other (please specify)
>
> Because the question as worded assumes far too much shared context and
> understanding in what we were looking for. We all understood what we
> wanted, but didn't unroll that in a specific enough way in the user survey.
Well said, Sean. The above is a great set of questions to ask in the
next user survey (or sooner?).
Best,
-jay
More information about the Openstack
mailing list