My biggest problem with
Google's
GoogleTalk Jabber service is that it treats a bare
e-mail address and a bare
Jabber ID as synonymous, disregarding the protocol schema and that the two inherently mean different things.
In many cases this might be a
safe, albeit erroneous, assumption. If a provider is offering both Jabber and
electronic mail services to the same set of people, it would be logical to make the mapping similar for convenience. Other times, the domain used would be different, and the assumption would result simply in messages getting dropped; for example an e-mail to user@jabber.example.com (instead of the correct user@mail.example.com) would be safe, in that eventually the sender should get an indication of the error.
However, there are cases where the two services would be running with the same domain, but for a different set of users. I once had a Jabber account at a service provider where an employee owned the same address for e-mail--the latter was a little irked that Google kept sending him messages that were targeted for me. This could be a situation where
bad things could happen, such as the wrong person getting private mail, and without any indication provided to either party.
The ridiculous thing, is that there is a way of
publishing contact information in Jabber, and for a system to pull the correct e-mail contact information from this data store. One would expect that any
major deployment would do the right thing here to save their users embarrassment. However, it is easy to err on the side of laziness in this situation. This is one of the issues I faced when writing
JabberAway. It's more complex to have a correctly done set-up shoe-horned into an
off-the-shelf system than to do
the wrong thing. However Google has no excuse. They have the money and man power to do the right thing, and wrote their system from scratch. Any mistakes or contrivances in their system are solely of their own poor design--and they should know better.