[Standards] IQ set semantics

[Peter Saint-Andre]

Tomasz Sterna wrote:
> Hi.
> 
> I was looking for an information whether the IQ-set sent on ones behalf
> (to change its own settings) may or may not have the "to" attribute and
> how should it be processed by the server.

We had some discussion about this a while back, and I tried to 
incorporate the results into the bis drafts. But maybe I didn't do a 
good job. :)

> http://www.xmpp.org/internet-drafts/draft-saintandre-rfc3920bis-04.html#stanzas-attributes-to-c2s
> "2. A stanza sent from a client to a server for direct processing by the
> server (e.g., presence sent to the server for broadcasting to other
> entities) SHOULD NOT possess a 'to' attribute."
> 
> It says that it SHOULD NOT posses a 'to' attribute. 

I think that one should be MUST NOT. Otherwise, how will the server know 
the presence is intended for broadcasting?

> But what to do if it
> does? Let's search further...
> 
> http://www.xmpp.org/internet-drafts/draft-saintandre-rfc3920bis-04.html#rules-local-node
> "3. If the JID is of the form <node at domain> and there exists at least
> one connected resource for the node, the recipient's server SHOULD
> deliver the stanza to at least one of the connected resources if the
> stanza is a message or presence stanza and SHOULD handle it directly on
> behalf of the node if the stanza is an IQ stanza."
> 
> OK. This says that the server should process the IQ on behalf of the
> "to" entity instead of forwarding to the client for processing.
> But I still am not sure what to do if the "to" is client's own bare JID.
> 
> The real life case is XEP-0054:
> "A user may update his or her vCard by sending an IQ of type "set" to
> the server, following the format in the previous use case.
> If a user attempts to perform an IQ set on another user's vCard, the
> server MUST return a 403 "Forbidden" error."
> 
> The "previous use case" does not contain a "to" attribute, so jabberd2
> implementation checks whether it is not set. If it is - returns
> forbidden-error.

The phrase "send an IQ-set to the server" is ambiguous. It could mean:

1. send an IQ-set over the stream you have established with your server 
but don't include a 'to' attribute"

or

2. send an IQ-set with to='yourserver.tld'

I think here (1) is intended but I can clean that up.

> Logic says that for IQs if "to" attribute is set to ones bare JID it
> should be handled like "to" was not set. But I cannot find a rule that
> says so in our protocol documentation.

I agree that we need to explicitly state that rule. I'll work it into 
the stanza handling rules.

> "In the face of ambiguity, refuse the temptation to guess." - PEP-0020

Good advice. :)

Peter

-- 
Peter Saint-Andre
https://stpeter.im/

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 7338 bytes
Desc: S/MIME Cryptographic Signature
Url : http://mail.jabber.org/pipermail/standards/attachments/20080114/1250fe3a/attachment.bin