- From: Xiaoshu Wang <wangxiao@musc.edu>
- Date: Mon, 24 Mar 2008 01:30:27 +0000
- To: "Roy T. Fielding" <fielding@gbiv.com>
- CC: Jonathan Rees <jar@creativecommons.org>, "www-tag@w3.org WG" <www-tag@w3.org>
Roy T. Fielding wrote: > > On Mar 20, 2008, at 5:41 PM, Jonathan Rees wrote: >> A URI manipulation convention such as Alan Ruttenberg's idea from >> last summer [1] would work for Aunt Tillie since it doesn't require >> any header or status code magic. Another solution would be a central >> registry or a set of registries. (I'm not saying these ideas aren't >> without faults.) What would you suggest? > > Google. Or SemanticGoogle, if you prefer. > > Don't kid yourself into thinking that anyone other than experts > will ever be interested in managing metadata. The 303 and Link > solutions are for Semantic Web experts to care about, content > management software to implement, and "online librarians" to > manipulate. Normal folks who just want a page on the Web do > not need any of this nonsense -- they don't care if links are > ambiguous and never will. Limiting the solution space to things > that those non-users understand is absurd, for the same reason > that Porsche doesn't design cars to be driven by five year olds. Ditto. Make it generic, make it practical, and make it useful. Make clear and useful definitions. If they can not be clearly defined, make them less restrictive but over-restrictive. That is the approach to do things. I think what has happened in the software engineer field has showed us plenty of example. Such as SOAP vs. REST and EJB vs. POJO. People who needs to deal with corner cases will always find a way to deal with them because they know exactly when and where they will need it. But don't design a system to cover corner cases. That will make the system too complicated to use. Xiaoshu
Received on Monday, 24 March 2008 01:31:16 UTC