How secure is quantum cryptography?

In 1991 scientists have for the first time demonstrated practical quantum cryptography (QC) and given a proof of its unconditional security against two most obvious attacking strategies. In the next decade several groups of scientists have made more stringent security proofs and it was soon widely accepted that quantum cryptographic protocol BB84 and its variants are the only known protocols that allow unconditionally secure key growth. Moreover, QC was able to offer unique possibility that any eavesdropping activity could be detected by legitimate parties. In 2004 Swiss spin-off IdQuantique has shown the world's first commercial quantum key distribution system Clavis, followed soon by the QPN system of American start-up Magiq. For almost 7 years nobody suspected that there might be something terribly wrong with these systems. Then, at the end of 2010, a group of Norwegian scientists experimenting with attacks on photon detectors constructed tailored attacks to both commercial systems which were so successful that the plaintext was recovered in full (100%) and moreover the legitimate parties could not detect any suspicious activity ! This raised the question: where does it leave us with quantum cryptography, is it secure or not? What about the security proofs, were they wrong? Our conclusion is that security proofs were correct except that they proved security of something else - not the machines that were cracked. At present state of the art it is not clear how or is it even possible to build a provable QC machine.