Identifiers#

Daft's SQL identifiers are case-sensitive by default, but support a case-insensitive and a case-normalize mode. For both the case-insensitive and case-sensitive modes, identifiers are case-preserved and are matched based upon the mode. For the case-normalize mode, unquoted (regular) identifiers are normalized to lowercase and double-quoted (delimited) identifiers are case-preserved. You can configure these modes via the SessionOptions when creating a session. These modes apply when resolving attached catalogs, attached tables, and columns via the session.

It is currently not feasible to ensure consistent casing semantics across catalogs, so we recommend using the different modes to find which best fits your preferences and current systems. When working across multiple systems, using all lowercase names for namespace and tables with identifier_mode = 'normalize' provides the most consistent experience.

Syntax#

1
2

-- regular identifier
abc

1
2

-- delimited identifier
"abc"

1
2

-- qualified identifier
abc.xyz

1
2

-- qualified identifier with mixed parts
a."b".c

1
2

-- delimited identifier with special characters
SELECT "🍺" FROM "🍻"

Rules

• Identifiers may be unquoted (regular) or double-quoted (delimited).
• Identifiers must be double-quoted if the text is a keyword or the text contains special characters.
• Regular identifiers must start with either an alphabetic character or an underscore '_' character.
• Regular identifiers must contain only alphanumeric, '$' and '_' characters.

Modes#

Configuration#

The identifier mode is configured at the session level. Currently, Daft uses case-sensitive mode by default.