[Standards] dialback refactoring

Philipp Hancke fippo at goodadvice.pages.de
Tue May 5 14:16:43 UTC 2009


Matthew Wild wrote:
[snip]
>>> Also regarding errors, I just noticed the XEP states: "Queued stanzas
>>> MUST be bounced back to the respective senders with a
>>> <remote-server-timeout/> stanza error". Wouldn't
>>> remote-server-not-found make more sense?
>> remote-server-not-found implies that the remote server does not exist,
>> i.e. a dns lookup failed. I don't think it should be used when you
>> successfully established a tcp connection to that host.
>>
> 
> and saying you got a timeout is better? :)

A timeout implies "try again" - see the error descriptions and
associated types in the latest 3920bis draft.

> I view "remote-server-not-found" above TCP. If the remote server gave
> "host-unknown" for example then I wouldn't consider that a timeout,
> rather I really wasn't able to find the correct server to receive a
> particular stanza - the remote server was not found (despite
> successful TCP connection).

changing the error type might break things. At least if anybody uses
the error type (-:

Bouncing with remote-server-timeout on the originating server upon
receiving type=invalid is nonsense btw. This indicates a (huge) problem
on the originating server. internal-server-error seems most appropriate.

philipp



More information about the Standards mailing list