Normalizer Bench

S

Short: NFD fits in 32 UTF-16 code units. French and German are adjusted to take a substring that contains a non-ASCII character. (Long input contains the same information in each language instead of having a fixed UTF-16 length.)

L

Latin1.

U

Forced UTF-16 form for Latin1 languages. (One non-Latin1 character added to the string.)

W

Forced write: In the UTF-16 case, a singleton is prepended to force the normalizer to start writing from the start. In the Latin1 case, a character with a compatibility decomposition is prepended, since there are no singletons in Latin1. This means the effect is seen only in the K forms.

C

Forced copy: In the UTF-16 case, a singleton is appended to force the normalizer to make a copy even when normalizing from NFC to a C form or from NFD to a D form. In the Latin1 case, a character with a compatibility decomposition is appended, since there are no singletons in Latin1. This means the effect is seen only in the K form corresponding to the input C or D form.

Bench not started.