I think those words are unfortunate and should be changed.
This blog post is perhaps a clearer description of how I see the
relationship between PuSH and Evented APIs.
http://www.windley.com/archives/2011/12/apis_that_call_you.shtml
Something that doesn't come through in the spec doc very clearly (and
needs to be updated) is the idea that Evented APIs are almost always
one-to-one, meaning that the event is being raised on behalf of an
entity. Think of connecting your account at Fitbit with your account
at Withings. I want Withings to know when there's new Fitbit data.
--phil--
Phillip J. Windley, Ph.D.www.windley.comwww.kynetx.com=windley

On Dec 12, 5:19 pm, Ka-Ping Yee <[email protected]> wrote:
> Hi Sam and Phillip,
>
> I'm curious about the Evented APIs spec and am trying to understand what
> motivated it.  I noticed this answer to the FAQ about PubSubHubbub, but it
> didn't really help:
>
> The complexity of using a distribution hub doesn’t make sense for anything
> but large systems. PuSH was a way to reduce polling on the origin server,
> but it’s really a stop-gap for better evented systems.
>
> "PuSH is just a stop-gap" seems like a content-free answer.  I imagine you
> must have had more detailed reasons in mind; would you mind elaborating a
> bit on what you mean by "stop-gap" and "better"?
>
> Thanks,
>
> —Ka-Ping Yee

Reply via email to