Data Type Naming

This topic discusses data types in configurations for extension adapters and data sources.

TDV Types

For extension adapters, TDV supertypes are named in such a way that no leaf type and supertype have the same name, even when prefix characters like ~ and @ have been removed. For example, supertype ~date is now referred to as dates, to distinguish it from the leaf type date (formerly denoted @date). The new type names appear in YAML configuration files and in Studio.

The table shows the source hierarchy of supertypes (in bold) and leaf types. The right column lists the names that are used for extension adapters.

Source (capabilities files for built-in adapters)				Target (YAML files)
`~any`				`any`
	`~number`			`numbers`
		`~whole_number`		`integers`
			`@bit`	`bit`
			`@integer`	`integer`
			`@smallint`	`smalling`
			`@tinyint`	`tinyint`
			`@bigint`	`bigint`
		`~floating_point`		`floats`
			`@float`	`float`
			`@real`	`real`
			`@double`	`double`
			`@decimal`	`decimal`
			`@numeric`	`numeric`
	`~string`			`strings`
		`@char`		`char`
		`@varchar`		`varchar`
		`@longvarchar`		`longvarchar`
	`~binary`			`binaries`
		`@binary`		`binary`
		`@varbinary`		`varbinary`
	`~date`			`dates`
		`@date`		`date`
		`@time`		`time`
		`@timestamp`		`timestamp`
	`~lob`			`lobs`
		`@clob`		`clob`
		`@blob`		`blob`
	`@boolean`			`boolean`
	`@null`			`null`
	`~interval_year`			`year_intervals`
		`@interval_year_to_month`		`interval_year_to_month`
		`@interval_year`		`interval_year`
		`@interval_month`		`interval_month`
	`~interval_day`			`day_intervals`
		`@interval_day`		`interval_day`
		`@interval_hour`		`interval_hour`
		`@interval_minute`		`interval_minute`
		`@interval_second`		`interval_second`
		`@interval_day_to_hour`		`interval_day_to_hour`
		`@interval_day_to_minute`		`interval_day_to_minute`
		`@interval_day_to_second`		`interval_day_to_second`
		`@interval_hour_to_minute`		`interval_hour_to_minute`
		`@interval_hour_to_second`		`interval_hour_to_second`
		`@interval_minute_to_second`		`interval_minute_to_second`
	`@xml`			`xml`

All Types

In extension adapter configurations and data source overrides, TDV and native data type names appear in these areas:

•

Data type mappings

•

Clause support

•

Operator mappings

•

Function mappings

Native and TDV types often have the same name—for example, TDV’s date and Oracle's date. In capabilities files for built-in adapters, the two are differentiated using prefixes. A TDV supertype name would start with ~ (for example, ~date), a TDV leaf type would begin with @
(for example, @date) and a native type would have no prefix (for example, Oracle’s native type called date).

For extension adapters, the @ and ~ prefixes are no longer used. Whole-word prefixes, followed by a dot, now distinguish between native and TDV types:

•

cis.date—TDV leaf type

•

cis.dates—TDV super type

•

native.date—Oracle’s native type

It is recommended that you use the prefixes only in those parts of the configuration file where both TDV and native types can appear; that is, function and operator mappings. For instance, the same function can list both native and TDV types in the same argument combination.

Because data type mappings and clause support specifically request either TDV or native types, no ambiguity can occur and so no prefixes are required:

•

The data type mapping configuration section specifies TDV-to-native mappings and native-to-TDV mappings separately.

•

A configuration item describing a particular clause specifies unsupported native types and TDV native types separately.

For these sections, prefixes are unnecessary and discouraged for the sake of brevity.

Here are examples of how to refer to types for data type mappings and clauses:

•

date—TDV leaf type

•

dates—TDV supertype

•

date—Oracle's native type

Defining Functions in YAML

If you want to define a function as STRICT, add a new function entry in YAML, for example:

ascii.strict:

      display_name: ASCII

      native_expression:

      arguments:

        combination1:

          arguments_in_order:

          - cis.numbers, cis.strings

Notice that just like its traditional capability counterpart, the function mapping's native_expression property is left as blank.

After that if the query has strict option enabled, then the function will not be pushed down:

SELECT {option strict}

ascii(rid) a

FROM