Encodings · TNID

An encoding maps one part of the 128-bit value to characters of the TNID string, <name>.<data>. The spec defines two, one per part; UUID hex, bytes and integers are representations, not encodings.

Encoding	Encodes	Alphabet / width	Why this one
Name	20 name bits ↔ the 1–4 characters before the `.`	null, `0`–`4`, `a`–`z`; 5 bits	A human-chosen, readable label: 32 symbols fit 4 characters into 20 bits
Data	102 data bits (TNID variant + payload) ↔ the 17 characters after the `.`	`-`, `0`–`9`, `A`–`Z`, `_`, `a`–`z`; 6 bits	Arbitrary bits, packed densely: 102 = 17 × 6, no padding

Composition

user.Br2flcNDfF6LYICnT = d6157337-0ebc-8686-83ab-4075a34cdcde:

Part	In the value	In the string	Encoding
Name	20 bits, `d6157`	`user`	Name
Separator		`.`
Data	102 bits: TNID variant + payload	`Br2flcNDfF6LYICnT`	Data
UUID version, variant	6 constant bits	Omitted

Sort order matches bit order

Value order is ASCII order in both alphabets, so sorting strings byte by byte sorts the bits. A TNID string therefore sorts like its 128-bit value.

URL safety

Both alphabets and the . are unreserved characters (RFC 3986 §2.3): no percent-encoding in URL paths, queries or fragments.