Sorry, this was way back, but caught my eye again.
At 11:39 05/01/27, Sam Ruby wrote: > >Martin Duerst wrote: >> At 01:51 05/01/26, Asbj n Ulsberg wrote: >> > >> >On Wed, 12 Jan 2005 16:54:27 -0500, Sam Ruby <[EMAIL PROTECTED]> >> >wrote: >> > >> >>> 2. Why MUST a feed point to an alternate version. [...] >> >> >> >> I'm -1 on removing this restriction. >> > >> >I thought we came to a sort of consensus that the link should be optional. >> >Or was that only for atom:entry? Anyway, I think both of them should be >> >optional. That is, I disagree with you, Sam. >> I agree with Asbjoern. Regards, Martin. > >There is consensus that atom:link is not required for atom:entries which >contain content. That consensus has been reflected in the most recent >drafts. That is not the question referred to above.
Understood.
>There are now, by some counts, ten versions of formats that call >themselves RSS. Every last one of then has a required channel/link. >Every last one of them. > >Relaxing a restriction requires consumers to handle more cases.
How difficult can it possibly be, in this specific case?
>My >expectation is that given limited demand for this feature, the more >likely scenario is that consumers will either produce unexpected results >or outright fail for feeds without this data.
What do they need it for in the first place?
>Because of this, I would like to request that there be a compelling use >case be found which for feeds for which there can not be a atom:link >defined.
Is there a compelling use case (not "it has always been that way", which doesn't count as an use case) for requiring it?
The use case for not having it is pretty straightforward: Just a feed. It's that simple. Feeds can live on their own. They don't need alternate Web pages, and if they don't have have them, it's just straightforward design to not require such an element. Everything else is just cheating ourselves.
>Note atom:link is defined as a URI. While most examples that >we have seen use the HTTP scheme, this is not a requirement. Given that >this is not a requirement, and given that existing RSS producers have >come out in mass demanding that this restriction be lifted from RSS, I >can only conclude that the burden seems rather low.
So you are saying that
<atom:feed>
<atom:head>
<atom:link rel="alternate"
href="noscheme://nonsense.example.org/this/here/to/keep/Sam/happy">
...is better than just leaving that <link> out? Do we want the above nonsense to go into the RDF model, and so on? Isn't it much more straightforward to model the format after actual and potential realities, rather than to have to cheat?
Regards, Martin.
