Loading ...
Sorry, an error occurred while loading the content.
 

Re: [rest-discuss] Media Type & Data Format Best Practice

Expand Messages
  • Jan Algermissen
    Hi Phil, ... I d say the 80/20 rule is a good guide: design your media type in a way that makes it possible to solve 80% of the envisioned problems at 20% of
    Message 1 of 4 , Sep 27, 2006
      Hi Phil,

      On Sep 27, 2006, at 1:09 PM, Phil Ruelle wrote:
      > Where do you draw the line between
      > generic and specific such that your type is not meaningless but nor
      > is is it so specific as to limit re-use?
      I'd say the 80/20 rule is a good guide: design your media type in a
      way that
      makes it possible to solve 80% of the envisioned problems at 20% of the
      effort and you get a chance that your type attracts interest and gains
      enough network effect that it eventually becomes a standard of some
      sort.

      If that happens you gain all the benefits of being able to use what
      most likely started in order to serve some specific intra-
      organisational purpose
      in communicating with your clients or partners with standadized means.
      Note that meanwhile you might have been cought by network effects of
      a 'competing'
      type yourself causing you to adapt your system and abandon *your*
      format. The
      outcome would be the same - more standardization.

      I guess this is where the benefit of true document style integration
      really
      is found: the above can happen (and in fact often did) if the
      interface is
      already uniform and we only have to deal with the payloads. I doubt that
      a similar (decentralized) process of 'natural selection' (excuse the
      Darwinistic touch :-) can ever lead to API standardization as WS-*
      proponents seem to think.

      Jan
      >
      > And on a related note what strategies have people employed when
      > defining data formats to make them extensible? Judicious use of 'any'
      > and 'choice' elements in XSDs lead to a horrible mess (IMHO). Are
      > people generally using Relax NG and Schematron to define data
      > structures? Of course this assumes that people are using XML ...
      >
      > I'd be extremely interested to hear any and all opinions on the
      > subject(s).
      >
      > Cheers,
      > Phil
      >
      >
      >
    Your message has been successfully submitted and would be delivered to recipients shortly.