Graham wrote:
The "Implementor's Guide" may or may not ever exist and may or may not be read by anyone, so that isn't an option.
The WG has tended to punt assuming on a Fabled Implementors Guide. Why is that punt not acceptable now?
Myself and Sam, and (I think) Tim and Antone, would all like the actual spec to make an actual recommendation that entries without enclosed content include a textual summary, if at all possible.
Bray voiced concerns about SHOULD; I suggested MAY. No further discussion occurred. What might be wrong with a MAY requirement in this case?
Note no one wants to ban title- only feeds if they come from title-only resources.
I've claimed here before that kind of reasoning is the same kind of reasoning that lead us to reject 3 dates. It's not something we can manage. What is different here?
[Also, the nature of resources ensures this is outside our scope; unless we're breaking with web arch, we deal squarely with representations; granted I may be reading too much into what you mean by resource, but I'm assuming its usage is technical until told otherwise.]
What language would you find acceptable that covers these bases?
Language that doesn't result in confusion. Sam's text pace makes title-only feeds bad practice. I'm not satisfied the text pace has adequate language - the rationale presented is weak, the SHOULD requirement level isn't entirely appropriate, the idea of deliberately issuing mixed messages from the spec is unsatisfactory. I've pushed back on this pace a few times now and haven't seen any good answers as a result of that. At some point someone really ought to provide a clear explanation of its value. What have I already said that you think is wrong so that I should change my position?
cheers Bill
