Homographs: Latin vs Cyrillic lookalikes and what we do about them

Typographer's warning. In many fonts these pairs are visually identical: Latin a and Cyrillic а, Latin e and Cyrillic е, Latin o and Cyrillic о, Latin p and Cyrillic р, Latin c and Cyrillic с.

That means ipoteka.yu typed in Latin could be confused with a name mixing Cyrillic letters. Because we proudly support both scripts, this is not a bug to ignore, it is a design task.

My proposals:

• A name must use one script only (plus digits and hyphen).
• The registry stores a skeleton form of each name; two names with the same skeleton are considered conflicting.
• The client shows the script of a name next to it, and highlights mixed-script requests as blocked.

Please poke holes.

Подржавам. Ћирилица и латиница су једнако наше. Једно писмо по имену је разумно. Али молим да се не третира ћирилица као 'опасно' писмо, само као друго.

@cirilica_nikola complete agreement. The wording in the policy must be symmetrical. Neither script is the suspect.

Skeleton approach is standard and works. Make sure the mapping table is published and versioned, and test with the Serbian and Macedonian specific letters too, not just the lookalikes.

I will create test vectors: pairs that must conflict and pairs that must not (for example č and ć are different letters, never collapse them to c).

And ѓ, ќ, ѕ in Macedonian must not be flattened either. A skeleton for confusability, not a destructive normalisation of the real name.

Exactly: display the name as the owner wrote it, compare by skeleton. I will write a one page visual guide with fonts that make the differences visible.