i foresee the Trezor replacing the offline wallet and its pc and being more secure since the offline wallets can be susceptible to a USB malware attack. as far as i know, there is no way to get privkeys off a Trezor so in that sense it is safer. i would hope you could just plug it into your online pc to sign tx's from the online Armory watching only wallets. we'll have to see what Armory comes up with.
OK thanks. So you are using the Trezor as a full replacement for the offline secure storage. And with the planned Armory support will continue to use online Armory watching only wallets as your wallet, but use the Trezor for signing.
That is neat. You keep the Armory front-end, while using a different back-end security platform for signing.
Personally for me I get worried about using any form of hardware that you have to plug into an online compute as the signing mechanism, it's just my preference but I prefer open-source code in machines that have never touched a network.
i don't understand how or why you are using HD seeds since Armory currently only generates linear deterministic keypairs, not hierarchical. i also wouldn't bother exporting privkeys from the offline Armory wallet since you can just make a HD seed backup of your Mycelium wallet that can be restored in case of theft or loss while avoiding excessive and unnecessary exposure to USB attacks from accessing the offline wallet.
Using standard Armory backups. HD seed is just a general term for any wallet where a single seed can be used to recover any number of addresses. Armory before has advertised itself as being the first "Hierarchical Deterministic Wallet", that is all I was referring too. Understand Armory uses a different type than BIP0032 today.
i wonder if another reason to avoid exporting Armory linear deterministic privkeys is if you reveal that privkey on your phone along with perhaps the master public key, can ALL your privkeys be determined similar to how it can be done with an xpub+childprivkey in HD wallets?
molecular would probably know.
I do not believe that in any deterministic wallet knowing the private key of any one address provides knowledge of other addresses. When I last looked into this a couple years ago that was my takeaway. If this is not true then that method is broken IMHO. But either way in my case I'd be fine since I use a separate wallet chain for those addresses, only one is nominally funded at a time, so have at it.
molecular, are there issues here?