Identifiers

Naming restrictions

• The first character of an identifier must be an ASCII letter (e.g., a-z and A-Z) or an underscore (_).

• The remaining characters of an identifier must be ASCII letters (e.g., a-z and A-Z), underscores (_), ASCII digits (0-9), or dollar signs ($).

• Non-ASCII characters in unquoted identifiers are not allowed.

• You can circumvent any above rules by double-quoting the identifier (e.g., "5_source"). All characters inside a quoted identifier are taken literally, except double-quotes must be escaped by writing two adjacent double-quotes, as in (e.g., "two""quotes").

• In an expression, certain names are interpreted as builtins rather than column names. For example:

  Names interpreted as builtins

  SELECT user; -- This is the builtin `user`.
  
  SELECT user, avatar FROM t; -- This is also the builtin `user`, rather than a column from the table `t`.

  Several such names require special attention, including user, current_timestamp, current_schema, current_role, current_user, and session_user. To avoid such issues, you can either avoid naming a column with these words, or qualify it with the table name (as shown in the example below) when such ambiguity happens.

  Solution to avoid naming conflicts

  SELECT t.user, avatar FROM t; -- Qualify it with `t.` to select the column rather than the builtin.

Case sensitivity

Identifiers are case-insensitive. It means wave, WAVE, and wAve are the same identifier in RisingWave. This can cause issues when column names come from data sources that do support case-sensitive names, such as Avro-formatted sources or CSV headers.

To avoid conflicts, double-quote all field names (e.g., "field_name") when working with case-sensitive sources.

RisingWave processes unquoted identifers as in lower cases. If you create a table named WAVE, it will display as wave when you choose to list all tables. You can reference it by wave, WAVE, or a combination of upper- and lower cases in SQL statements.