Re: ETags?

On Jan 19, 2005, at 3:15 PM, Geoffrey M Clemm wrote:
> I heartily agree with everything Roy says here.
> Roy: WRT your concern about the statement about live properties,
> would it be OK instead to say:
>     2.6  PROPFIND and Bindings
>      Consistent with [RFC2518] the value of a dead property of a
>      given resource MUST be independent of path to that resource
>      submitted to PROPFIND.  The value of a live property SHOULD
>      be independent of the path submitted to PROPFIND, unless
>      the definition of the property explicitly states otherwise.

I don't like meaningless SHOULDs.  How is an implementation supposed
to test compliance with such a requirement?  I would prefer that the
specification say nothing about the value of a live property,
since live properties are (by definition) not controlled by the
client and thus not subject to interoperability constraints aside
from whatever may be in their definition.  The only thing that can
be legitimately said is that

   "The value of a live property MUST comply with the definition
    of that property."

which is, of course, a completely vacuous statement and not
subject to the bindings specification.


Received on Wednesday, 19 January 2005 23:38:34 UTC