As part of Sun's encrypting firewall project, I conducted a code review of the Crypto iButton with Dallas Semiconductor's development team. Got some tips from Whit Diffie before going there and had Tsutomu Shimomura on speaker phone during the review. We were quite satisfied with their physical and code implementation. A few code-paranoia suggestions were made and were subsequently implemented. But the opinion was that it was secure even without those changes.

The Java iButton was developed by the same team on an improved version of their hardware. I would expect that it would have the same quality of implementation.

I haven't talked to the DalSemi folks much since the merger, but I regard them as one of the best vendors I had ever worked with.

Ben