On 8/16/11 3:35 AM, Dave Cridland wrote:
> On Mon Aug 15 22:57:59 2011, Peter Saint-Andre wrote:
>> > 1) I don't think the new elements need to have a 1:1 mapping with
>> status
>> > codes, and the way in which the registrar considerations is phrased
>> > implies this to be the case. I think we want a new registry, which
>> > includes "related status codes".
>>
>> I think we want to define new status condition elements for each of the
>> old status codes, but that we want to leave open the possibility of
>> defining new status conditions as well (which would be represented only
>> via XML elements, not code numbers).
>>
>>
> I think we want to have at least one status condition for any given
> status code, but I don't think we need or want a 1:1 mapping.

While we're at it we might reconsider some of the existing conditions
(and not map them to official elements -- they could be extensions for
particular implementations).

>> > 2) We also need to consider that many clients only handle one status
>> > code.
>>
>> Which one do they handle?
>>
>>
> I mean they will only process one in a set. Sometimes they ignore all
> but the first. Sometimes they ignore all but the last.

Um, isn't that called a bug?

>> > This means that even if there are multiple defined conditions, all
>> > the status codes possible may not be present. In some XEP, somewhere, a
>> > discussion on which status code to pick when there are multiple
>> > applicable would be useful.
>>
>> Right. Similar to XEP-0086.
>>
>>
> I think more "If you want to say <room-created/><self-presence/>, then
> you only want to mention status code 201", as a specific example
> required for interop.
> 
> 
>> > 3) I'd like application-specific elements to be possible within the
>> > defined conditions as well. This is to encourage implementors to
>> > specialize, rather than replace, conditions.
>>
>> Section 2 states in part:
>>
>>     ... the <conditions/> element MAY contain one or more condition
>>     elements defined in this document (each of which MAY contain a
>>     human-readable <text/> element and MAY contain an application-
>>     specific condition element)
>>
>> So that's already covered.
> 
> Oh, right, yes. I entirely missed both this and the structure above it
> which makes this clear. Damn it, it's not in the examples!

Yeah, I wrote the proposal in a hurry. More examples are always good.

/psa



Reply via email to