Re: [weirds] Ted Lemon's Discuss on draft-ietf-weirds-rdap-query-16: (with DISCUSS and COMMENT)

[Ted Lemon <[e-mail]>]

On Oct 29, 2014, at 8:32 PM, Pete Resnick <[e-mail]> wrote:

> On 10/29/14 5:09 PM, Ted Lemon wrote:
>> Why not specify what the right thing is for the client to do as a SHOULD, and _then_ explain what might break if it doesn't, rather than just sort of talking about it and hoping for the best?
>>   
> 
> But that's exactly what the second full paragraph of 3.1.3 says now: SHOULD be either U-Labels or A-Labels, might break if not (see section 6.1), here are reasons you might see brokenness, and servers can do what they will to make things right.

Right, fair enough.   But what's bothering me is the text that follows that.   Why isn't it a SHOULD that the server should convert U-labels to A-labels?

>> Also, U-labels are case-sensitive.   So the semantic difference you're describing doesn't actually exist.
> 
> If servers could count on only getting real U-labels, of course. And if servers could simply reject crappy input out of hand, life would be much simpler. But AFAIU this is a case in which some servers (which serve particular constituencies) have to be a bit liberal in what they accept. So I think we have to give some information in the document about what to do when you get "interesting" input.

What's a "real U-label?"   If it's unicode, it's a U-label, or else it's garbage.   There is no third alternative.   If it's garbage, just reject it.   When would garbage be something that it would make sense to attempt to use, rather than rejecting?   I think never.