Dean suggested that for any of the common object languages that a key be
reified as a language object which could be wielded by any key savvy code
written in that language.
In addition for the common Keykos objects there would be object types that corresponded to the Keykos type.
For Java there would be a ‘SpaceBank’ class each instance of which would access a Keykos space bank via a Keykos key.
There are several issues
- Who is responsible for the bank key held by the Java instance being a real Keykos bank?
- Is this compaibble with a key cache?
- Is this compatible of running without a key cache, as most current objects do?
Here I consider language selection.