As far as I know the RESTful architecture was first introduced by Roy Thomas Fielding in his Ph.D. dissertation. What I didn't know was that he wrote it in 2000, quite a while ago. I guess that choosing the best protocol (or rather rules when it comes to REST) is about convenience. Right now I'm particularly thinking of projects such as CouchDB, I'm sure its creators must have had a good reason to build a RESTful interface rather than XML-RPC (or SOAP), perhaps because it was a more convenient choice for them, especially in terms of development. At the same time this makes developers' lives harder when it comes to keep compatibility with earlier versions of a RESTful API. As you said REST is not a protocol therefore you could easily break compatibility although someone might argue that CouchDB is still undergoing heavy development.
Thanks for your comment, you made a very good point.
As far as I know the RESTful
As far as I know the RESTful architecture was first introduced by Roy Thomas Fielding in his Ph.D. dissertation. What I didn't know was that he wrote it in 2000, quite a while ago. I guess that choosing the best protocol (or rather rules when it comes to REST) is about convenience. Right now I'm particularly thinking of projects such as CouchDB, I'm sure its creators must have had a good reason to build a RESTful interface rather than XML-RPC (or SOAP), perhaps because it was a more convenient choice for them, especially in terms of development. At the same time this makes developers' lives harder when it comes to keep compatibility with earlier versions of a RESTful API. As you said REST is not a protocol therefore you could easily break compatibility although someone might argue that CouchDB is still undergoing heavy development.
Thanks for your comment, you made a very good point.