When my department (mathematics) tries to hire or promote someone with publications in  selective computer science conferences (like FOCS), we explicitly point out in the file (which goes first to our own faculty and then to the higher administration) that acceptance of a paper in such a conference is equivalent to acceptance in a high-quality journal.  The equivalence is, by now, fairly well-known, but we don't want to take the risk that someone who doesn't know it will cause trouble.  

Turning to your more general question, I think breadth is generally viewed positively.  What would, of course, be a problem is if people in area X say "well, he's not all that great in X, but we're impressed that he also does Y" while the people in area Y say "well, he's not that great in Y, but we're impressed that he also does X."  Once these people talk to each other, "not so great" will win and "impressed" will disappear.