There's already a package and command-line tool available for manipulating bech32 encoding. Instead of incorporating this feature in the codebase, leading to more code to ship and maintain, could we not provide guidance to users in error messages or documentation?

indeed, bech32 works fine. I use it everyday. I was thinking about wrapping it under the hood of proposed key convert command. It might be superfluous as people are already accustomed to bench32 (myself included). Nevertheless, I would still go with adding the --hex functionality where missing and might be of use. For example, below

cardano-address key private --signing-key --hex < drep.xsk
cardano-address key hash --hex < .pay.xvk

works, but there are some places where it is missing, like

cardano-address key child
cardano-address key from-recovery-phrase
cardano-address key public

Here, we should add optional FormatType and get the same experience. That would be my proposal.

OK, so the idea is to make commands more consistent? Then why not remove --hex parameter and point the user towards bech32? This is less work, reduces the amount of code we maintain, improves consistency of commands, and aligns better with unix philosophy (do one thing well, compose through input/output).

hmm, having bech32 available and working perfectly, I start to incline towards cleaning and slimming cardano-addresses out of --hex . Very "linux in mind compability" proposal 💯 I am happy to handle that 👍