- From: Roy T. Fielding <fielding@gbiv.com>
- Date: Tue, 20 Dec 2011 18:27:39 -0800
- To: Kevin Smith <kevsmith@adobe.com>
- Cc: "<public-tracking@w3.org> (public-tracking@w3.org)" <public-tracking@w3.org>
On Dec 20, 2011, at 3:35 PM, Kevin Smith wrote: > I prefer to leave syntax open to be extended, but not actually write in unused syntax until its needed. Future-proof means don't paint yourself into a corner - make it possible to easily add stuff in the future. It does not mean - try to guess what we will want in the future and add it now. I don't have to guess what DNT: 0 means. The question was only if we'd ever encourage it to be sent, which is a decision that can be left to others if and only if deployed recipients will understand what it means, and the only way that is going to happen is if we include the definition now (before it is used but not before the definition is needed within soon-to-be-deployed software). After all, Adobe wouldn't want to fail on a DNT: 0 signal, even if we don't know why the user chose to send it. In any case, the difference between REST as a design style and most other information systems design is precisely the concern for future, often unanticipated, reuse. I know it contradicts some of the long-bearded principles of project management, but I think I can safely say that REST has already won that argument for network-based systems (just as McIlroy's pipe-and-filter design won it for command-line systems long before REST). It is better to define a uniform interface from the get-go and let other people decide what parts to use. ....Roy
Received on Wednesday, 21 December 2011 02:28:09 UTC