Explicit Character Exclusions in the Encoding

[kyzer-davis]

• Discussion Thread
• Some libraries exclude specific alphabetic characters: Oo, Ii and Ll, sometimes Uu?
• These characters help human readability since I/l can be confused with 1 and O can be confused with 0 while the U character can create accidental obscenities in the encoding.
  • Also scenarios where "v==u" and "2==z" as seen in z-base-32
• While usually specific to the alphabetic characters, there may be gaps in the numeric characters available too.
• Note that qualifier is "Y" if an encoding goes out of their way to exclude a character. e.g Base32hex does not use the full A-Z alphabet because it is not required, not for some other specific purpose as noted above.

[sergeyprokhorenko]

It is better to sacrifice the digits 0 and 1 than the letters O and I, and exclude lower case. A new proposed alphabet based on Crockford's alphabet is 23456789ABCDEFGHIJKMNOPQRSTVWXYZ, uppercase only. See also Base32-Base, Base32-NCName, https://brandur.org/fragments/base32-slugs

The reasons are:

1. The digit 0 is somtimes pronounced as "oh", especially in British English in phone numbers.
2. Leading zeros can be perceived as unsafe, since they are omitted from numbers.
3. Too many decoded symbols for 0 and 1.
4. Equal-height characters make it easier to see the boundaries of the UUID and identify the UUID, as in the case of abbreviations.