I think we are in the third case:
Maybe useful but none has yet proposed it with convincing use cases?
It is true that the keys must be given from the outside but so are the signatures so if one needs to combine signature checking and address-based authentication it is always possible to replace all occurrences of
pair key signature and use the
HASH_KEY instruction (possibly followed by
ADDRESS) to check that a given
key and a given
What we currently cannot simulate however is the ability to treat revealed accounts differently from not-yet-revealed ones and I don't see a use-case for this.
I don't see how having access the revealed keys could endanger the security.
SENDER authentication solves many basic use cases but is not enough for some applications (multi-signature and fee-less transactions in particular) which rely on signature-based authentication instead. Finally note that another authentication feature named Tickets is being developed and might be proposed in a 007 protocol amendment.