Open Stack

On 2015-08-27 16:13:29 -0500 (-0500), Anne Gentle wrote:
[...]
> In order to serve API reference docs, we need a Pecan server on a cloud
> server of some sort. I'm guessing this server could be deployed with Puppet
> but I really don't know.
> 
> The specification for revising API docs [1] gives some detail about the
> fairy-slipper tool [2] being developed that gets us out of WADL and into
> JSON. The JSON needs a Python web framework to host the reference content.
> An example of the output is at [3].
[...]

Trying to summarize some of the subsequent discussion[1] which took
place last week in #openstack-infra: while not ruling out the
benefits of a separate server (particularly for interesting future
additions like an interactive API sandbox), there are still good
reasons to prefer processes which independently build static content
for upload instead of regenerating the same content on demand on the
server where it's hosted (since it can also be easily packaged and
redistributed if it has a known state).

An option worth looking into is whether fairy-slipper could be
pointed at running services in a devstack-gate job to generate that
JSON, and also perhaps whether the resulting JSON could be
pre-rendered into its final state prior to upload to the Web site or
if necessary integrated into a page template client-side with some
Javascript.

[1] <URL: http://eavesdrop.openstack.org/irclogs/%23openstack-infra/%23openstack-infra.2015-08-28.log.html#t2015-08-28T20:20:45 >
-- 
Jeremy Stanley