Mark Baker wrote: > > > On 6/5/07, Anil John <[EMAIL PROTECTED] <mailto:aniltj%40gmail.com>> wrote: > > So... Is GET supposed to be "safe" or not when it comes to REST? > > REST doesn't care. It doesn't even require a GET-like operation be > available. But GET is defined by HTTP to be safe, so it's most > definitely bad practice to attribute a state change on a GET to the > user.
One of the predominate design problems with HTTP URLs is the complexity of sent content vs URI indicated values. So, many people do build entire web applications treating GET like POST because they can indicate in the URI everything that they would otherwise have to tack onto PUT or POST with content that is more verbose or more work to manage on both ends. The real question is whether the interface is this set of GET operations, or another, more visible interface such as a web page or an applet that uses GET behind the scenes and doesn't advertise an URI structure as the interface. Gregg Wonderly
