[Standards] Zero-length nodes and resources

Peter Saint-Andre stpeter at stpeter.im
Fri Dec 5 04:58:29 UTC 2008

Matthew Wild wrote:
> On Fri, Dec 5, 2008 at 2:01 AM, Peter Saint-Andre <stpeter at stpeter.im> wrote:
>> Matthew Wild wrote:
>>> Hi,
>>> I got into a debate today over some unit tests I wrote, which
>>> explicitly checked that we handled JIDs such as "@example.com" and
>>> "user at example.com/" as invalid. However it seems the RFCs don't
>>> explicitly disallow either of these cases.
>>> How should they be handled? I can imagine that we currently have
>>> implementations both that declare them invalid, and those that handle
>>> zero-length as non-existent (almost certainly incorrect).
>> rfc3920bis disallows those cases -- this was a spec bug in RFC 3920.
> Thanks, that settles it.
> I glanced over the ABNF, I was looking for the requirement text in the
> node/resource section specifically, though I noticed where it was when
> I looked at the changes.

I will add some explicit text about "this identifier must not be of zero
length" to Sections 3.3 and 3.4 of rfc3920bis. Sorry for the confusion.


More information about the Standards mailing list