That's correct. There are often concerns about the scope of mustUnderstand and failures. One of the proposals in the past was that a mustUnderstand on an entry would not cause a feed aggregator to fail. The problem is what is a feed aggregator?
XML defines 2 conformance levels, well-formed and valid. Atom could easily do something similar, say a "feed" and an "entry" conformance level for mU. I had asked for some explanation of various profiles previously but didn't get any response. But even that definition can be complex and people are leery about defining any kind of processor/conformance/profile levels. Another approach is to split the mU into 2 levels: feed mUs and entry mUs. BTW, this is roughly why SOAP introduce the "role" attribute. But I haven't been able to clearly think through how soap roles could be used in combination with the various models of processors to handle mU. Cheers, Dave > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:owner-atom- > [EMAIL PROTECTED] On Behalf Of Walter Underwood > Sent: Friday, November 12, 2004 9:23 AM > To: Atom WG > Subject: Re: Published extensibility Paces > > > --On Thursday, November 11, 2004 11:32:32 PM +0000 Graham <[EMAIL PROTECTED]> > wrote: > > > > I have decided that Shrook understands all possible extensions but is > > consciously choosing not to display whatever they might contain. Thus I > > don't have to implement any of this nonsense. > > Extensions might not be separate displayable things. Maybe it is info > for a micropayment scheme. > > In general Shrook would have to implement it, but the implementation > is very simple. > > If the doc has MustUnderstand extensions, and the software doesn't > implement all of them, then the software must reject the entire document. > Not just the extensions, the whole thing. Compare it against the local > list, > then toss or pass the whole feed. > > I'm not arguing for or against MustUnderstand, just trying to keep > this accurate. > > wunder > -- > Walter Underwood > Principal Architect > Verity Ultraseek
