Re: Federation protocols

On Sat, Jun 1, 2013 at 9:26 PM, Andreas Kuckartz <A.Kuckartz@ping.de> wrote:

> Nick Jennings:
> > I say it's not really about the protocol(s), it's about our approach in
> > developing open alternatives for end-users.
>
> That is how I understand this approach:
>
> * Create some kind of glue which can be used to connect different kinds
> of FSW nodes
>
>
Either connect the nodes directly (more difficult) or abstract away the
protocol specifics from the client application, while still having the
interaction be essentially client-server (more easily accomplished).

Federating two server nodes together, from two separate projects, brings up
a lot of questions about what that actually means and in the end I think it
may only make sense for a very small subset of data from each of the
separate projects. (ie. sending messages back and forth), but not
"native-like" interaction between users of separate projects.



> * That reduces the amount of work required for the different FSW
> projects to collaborate
>
>
Yes, as long as their APIs and protocols are sound, and they allow
third-party interaction, then I think we can make things work.
Authentication is the big issue here, but solvable and there are
work-arounds.

Closing off an API is a big danger, though, sites like Twitter seem to be
closing off more and more. For open alternatives, this isn't really an
issue.



> * Make the software created by FSW projects more modular
>

Yes, I think it makes sense to abstract away as much of the protocol
specifics as early as possible. So, client-facing interfaces, although
perhaps originally designed by one project, could theoretically be used in
other contexts with similar types of data. For example. Developer A makes
an awesome twitter client, using an abstraction layer for any and all
protocol specifics. Let's say some new twitter-esque open alternatives
comes on the market. That awesome twitter client could easily leverage the
new site with very little change to the application.


Correct?
>

Yes I think you've summed it up more or less.