We say things like "ISO 4217 code, plus others in common use" in the spec, but we need to be much more formal about this. That helps us avoid problems such as we had with the currency mapping table, and gives users a strong sense of what to expect. I recommend that we have a formal definition of Unicode Currency ID and similar items. In all cases, we make it clear where these are based on other standards, and how they track them, but we then have a formal way for people to reference what we use, which may be are extensions of the base standards.
The /Unicode Currency Identifier/ is used for identifying currencies in software. It is based on 4217, and tracks it closely, but also adds some additional codes for currencies in common use. The values are always 3 ASCII letters, compared case-insensitively. They are listed explicitly in bcp47/currency.xml. For more information, see XXX.
and so on.
This should be done for each of the items in http://www.unicode.org/reports/tr35/#Key_Type_Definitions