Mark has an interesting post on the (lack of) requirements for humans in the face of GET.
No matter where you land on the REST spectrum, I think it's very easy to make the case for GET as an exceedingly common idiom in “machine facing” scenarios.
There are tons of precedents here outside of the web context:
1. COM Monikers. Very similar in both syntax and function to GET albeit in a proprietary programming system rather than a protocol.
2. OLE2's IDataObject.GetData. Again, a uniform mechanism for retriving representations (with content negotiation no less!). Very heavily used even today.
3. Database connection strings/SQL queries. Yet another string-based scheme for referencing data against a uniform (albiet plugable) access mechanism.
4. S-expressions. OK, this one is a bit of a stretch (especially if you deeply buy into the “Resource“ in REST). But again, uniform textual syntax + a uniform “access” mechanism (eval).
I've used the “URL as S-expr/HTTP GET as R-eval“ analogy more than once and its a facinating discussion to have, especially if you're open to alternative theories as to how the web actually works.
Posted
Mar 07 2007, 02:24 AM
by
don-box