Danny is correct - there is actually *no such thing as a double-spend* (check the entire blockchain history and you'll not find one).
There are only "double-spend attempts" which then the miners have to decide on *which* of the options to accept (but only one).
I think this depends on your definition of "double spend"
If you would define a double spend as the exact same input being spent more then once then you are correct as the network would reject any found block that contains anything like this.
If you would define a double spend as a TX in which the same input(s) are part of a transaction that would prevent a separate, previously propagated TX that someone though was going to be confirmed, then this is very much possible. For example, if send a TX with inputs (a), (b), and (c) to a retailer and then later are able to get a separate TX to confirm that contains input (c) (the TX would likely be to an address that you control) then this would be a double spend under this definition.