- From: Clemm, Geoff <gclemm@rational.com>
- Date: Thu, 21 Jun 2001 14:59:41 -0400
- To: ietf-dav-versioning@w3.org
From: Lisa Dusseault [mailto:lisa@xythos.com] I second what Jim's saying. Furthermore, I'd point out that server implementations differ. Servers may not implement all live properties or methods that a client expects. Some servers may add new, custom live properties. Does that change the type and make the client unable to confidently deal with the resource? So, how does a client interoperate with such a server? Either the server lies, and says it supports a given "resourcetype" when it only supports a subset of the methods and properties defined for that "resourcetype", or it makes up some new server-specific resourcetype for that resource. In the former case, an interoperable client assumes that the server is telling the truth, only to find that certain unimplemented methods fail and certain live properties are absent or have unexpected values. In the latter case, an interoperable client (that hasn't been coded specifically against that server) says "I don't know what that resource is", and doesn't try to access any methods or properties from that resource (since it doesn't recognize the value in the DAV:resourcetype property). Now suppose instead the client uses the values in DAV:supported-live-property-set and DAV:supported-method-set to determine if a given resource provides a set of services that it will use (note that this might be a subset of the defined properties and methods for a "resourcetype", since it might not use all the defined properties and methods). In this case, the client works properly against a server that either subsets or extends the supported methods and properties. So I applaud your argument, but point out that it leads to the opposite conclusion from the one you reached. So ask yourself if removing resource types in DeltaV improves simplicity. It does not. The simplicity argument is minor to irrelevant. The argument against DAV:resourcetype is that it is at best redundant, and at worst, as your scenario illustrates, it is even harmful since it decreases the likelihood that a client will interoperate with a wide range of servers. Cheers, Geoff
Received on Thursday, 21 June 2001 14:54:05 UTC