Unicode Sets


Certain fields have sets of characters (and strings) as values, called Unicode Sets. These have the following format in CLDR 44 or later:

In CLDR 43 and previous versions, a different format was used, one that require special "escapes" for certain characters and for strings. This caused problems for many people, and was replaced by the simpler format above.


In the info panel, a mouse hover over the non-winning values shows a comparison to the Winning value. The ➕ { } indicates that { and } are additions to the Winning value, and ➖ ‐ – … ' ‘ ’ " “ ” § @ * / & # † ′ ″ indicates that ➖, ‐. –. …. and so on are subtractions from the Winning value. That makes it much easier to see what the difference in the outcome would be.

The very last line shows an internal UnicodeSet format. You can normally ignore this. However, if you want more details about the characters you can copy the [...] from that line in the Info Panel and paste that into the Input box on UnicodeSet (and hit Show Set) to see more information about the characters, such as [!(),-.\:;?\[\]\{\}‑].

Exemplar Characters

The exemplar character sets contain the commonly used letters for a given modern form of a language. These are used for testing and for determining the appropriate repertoire of letters for various tasks, like choosing charset converters that can handle a given language. The term “letter” is interpreted broadly, and includes characters used to form words, such as 是 or 가. It should not include presentation forms, like U+FE90 ( ‎ﺐ‎ ) ARABIC LETTER BEH FINAL FORM, or isolated Jamo characters (for Hangul). 

There are different categories:

Examplar Examples

Parse Characters

These are sets of characters that are treated as equivalent in parsing. In the Code column you'll see a description of the characters with a sample in parentheses. For example, the following indicates that in date/time parsing, when someone types any of the characters in the Winning column, they should be treated as equivalent to ":". 

Note that if your language doesn't use any of these characters in date and times, the value doesn't really matter, and you can simply vote for the default value. For example, if a time is represented by "3.20" instead of "3:20", then it doesn't matter which characters are equivalent to ":".

Handling Warnings in Exemplar characters

There are two kinds of warnings you can get with Exemplar Characters. While these are categorized as warnings, every effort should be made to fix them.

A. A particular translated item contains characters that aren't in the exemplars.

For example:

Three possible solutions:

B. The exemplar characters shouldn't contain a particular character.

The standard characters shouldn't contain punctuation. They also should not contain symbols, unless those symbols are only used with the language's writing system (aka script). For example, the standard Bengali currency symbols should contain the Bengali Rupee mark (which is Bengali-only), but should not include the $ Dollar Sign (which is common across all scripts).