Suitable is subject to discussion, so I will try to stick to facts and the current (fast moving) state of implementation.
Geth for instance plans rudimentary ICAP support from version 1.4.1 onwards.
Confusion exists on actually why there has to be yet another standard of addresses checksum. Critics state the checksum encoded hex addresses are not immidiatly recognizable as being distinct from regular hex addresses and therefor the added layer of security only appears on a software level, not on a user level.
Therefore, implementing and using ICAP should be the most suitable option, since it's adding more security for the users and is currently the most distributed checksum enabled address format, even though not all clients support it. Yet.