Hi,

2017-03-09 17:20 GMT+01:00 Florian Schmaus <[email protected]>:
> That was not the motivation behind the namespace bump. I think it is
> possible that there are server implementations of carbons which do not
> restore the carbon state after resumption.

Maybe there are. Do you know of one in particular?

> And if we assume the
> existence of such implementations and that they are not broken, then we
> would need to bump the carbon NS in order to specify that they carbons
> state is restored after resumption if we want clients to exploit that.

I would argue that those implementations are in fact broken and need to be fixed

> I see people are arguing that its pretty clear what the state after a
> resumption is. I am not sure if this is the case.

Why is that? Why are you not sure about that?

>
> For example, clients *can not* assume that the CSI state is restored in
> every case.

I believe that it is clear that the CSI state is restored to what ever
state-request the server received last. The fact that the client does
not know what state-request was received by the server because csi
doesn't have an IQ to and isn't counted as a stanza might or might not
be a problem. That could be fixed by making CSI an IQ.
Again: The state on the server side is actually pretty clear. It is
what ever it was before. I agree however that the client might not
know the state.

> Also we don't restore the stream compression state after resumption (or
> do/should we?). So why is it clear that the state of other XEPs is restored?

Compression gets negotiated before SM. SM only resumes the inner
stream. You could raise a similar question with TLS. Does the TLS
state get resumed with SM? This question doesn't make sense. You could
resume a stream that was created over a TLS connection in a plain text
connection. Same goes for compression. If you want compression or TLS
for that matter in the new connection as well you negotiate that first
and then resume a stream.

cheers
Daniel
_______________________________________________
Standards mailing list
Info: https://mail.jabber.org/mailman/listinfo/standards
Unsubscribe: [email protected]
_______________________________________________

Reply via email to