Re: How much REST should your Web API get?

Jérôme Louvel <[email protected]> Fri, 3 May 2013 12:18:55 +0200
Newsgroups gmane.comp.web.services.rest
Message-ID <CAOsY8mSS+kNF02JyV7Jvo7Z8uw11fTuamuB+E=X0i1ctQ2_+ow@mail.gmail.com>
Excellent feed-back Robert.

I have entered several issues in the GitHub project to keep track of them
and ensure they are adressed in the more complete document to be published:
https://github.com/restlet/web-api-style/issues?state=open

Best regards,
Jerome


2013/5/2 Robert Brewer <[email protected]>

> mike amundsen wrote:
> > I'd like to see more work on using the "Taylor School" approach
> > to modeling software arch (Properties/Qualities, Requirements,
> > and Constraints)
>
> Me too. And that will only be useful if folks are educated on the
> differences. Therefore...
>
> In this case, the model needs to be fixed. "Mobility" is not a constraint,
> it's a requirement. Most of the bullet points in that section are
> requirements. "Off-line application mode" (I would rephrase it to
> "disconnected operation") would be the constraint. That constraint plus the
> cache constraint could lead to *an architecure* which syncs caches on
> reconnection, but that design decision is not part of the style.
>
> Under "Custom Interface", it would be very informative to lay out in
> detail which of the properties are being traded off when selecting explicit
> versioning. Not to evangelize them over REST or vice-versa, but to
> understand in which situations each is appropriate.
>
> Roy himself isn't a REST evangelist. He's an accomplished architect who
> has created multiple architectures using multiple styles, attempting to
> make them appropriate for the given requirements, both functional and
> non-functional (properties). We would all be wise to design our
> architectures with the same unbiased approach.
>
>
> Robert Brewer
> [email protected]
>