[mdx] draft-young-md-query new draft 06 posted

[Tom Scavo]

On Sat, Jan 14, 2017 at 6:53 AM, Ian Young <ian at iay.org.uk> wrote:
>
> On 13 Jan 2017, at 21:50, Tom Scavo <trscavo at gmail.com> wrote:
>
> Okay, so I think you're saying this is the *minimal* encoding necessary:
>
> http://metadata.example.org/service/entities/http:%2F%2Fexample.org%2Fidp
>
>> Yes.
>>
> but I assume full per-cent encoding will work as well:
>
>> Yes, over-encoding is fine in the sense that it should give the same result.

Then I think you need to say that:

1. The character '/' MUST be percent-encoded.

2. The space character MUST be encoded as '%20' and MUST NOT be
encoded as '+' as would be required in a query parameter.

3. Any other character MAY be percent-encoded.

> Over-encoding sometimes happens in practice because which encoding method to
> use can be a little confusing and people use the wrong one...

Indeed. It's a total mess in practice.

In practice, client software will expose one or more encoding
routines. The user needs to make a choice. That's why I'm suggesting
the catch-all third item be added to the spec. Without that, it's not
clear to the user whether a particular routine is acceptable.

> it just didn't
> seem like a good idea to propagate that kind of mistake, even in a
> (non-normative) example.

I agree but OTOH I'll suggest you also display

http://metadata.example.org/service/entities/http%3A%2F%2Fexample.org%2Fidp

and note that it is acceptable as well. That's the output of a full
percent encoder, which is the most likely routine one will find in
practice.

Finally, the spec doesn't say anything about what a conforming server
should do. That's not obvious. If it were, we wouldn't have to go to
such great lengths to describe client behavior.

Tom