Instructions

1. Field Types Overview

LookML fields are the building blocks of your data model. Each type serves a specific purpose in generating SQL.

2. The Role of sql Parameter

The sql parameter behaves differently strictly based on the field type.

Dimensions: The "What"

• Role: Defines the raw transformation of the column before any aggregation.
• SQL Context: The expression is placed directly into the GROUP BY clause.
• Input: Can reference table columns (${TABLE}.col), other dimensions (${dim}), or raw SQL functions.
• Example:

``lookml dimension: fullname { sql: CONCAT(${firstname}, ' ', ${lastname}) ;; } -- Generates: CONCAT(table.firstname, ' ', table.last_name) ``

Measures: The "How Much"

• Role: Defines the value to be aggregated or the calculation involving other aggregates.
• SQL Context: Puts the expression inside the aggregation function (e.g., SUM(sql)), or as a standalone calculation for type: number.
• Input:

For type: sum/avg/min/max: References dimensions or columns. For type: number: References other measures. For type: count: sql is ignored (always COUNT() or COUNT(primary_key)).

• Example:

``lookml measure: totalprofit { type: sum sql: ${saleprice} - ${cost} ;; } -- Generates: SUM(sale_price - cost) ``

Filters: The "Which"

• Role: Defines the condition logic, usually for Templated Filters used in Derived Tables or sqlalwayswhere.
• SQL Context: The sql parameter in a filter field is rarely used directly in modern LookML. Instead, the input to the filter is used in {% condition %} tags.
• Best Practice: Identify if you need a filter field or just a parameter + dimension.
• Example (Templated Filter):

``lookml filter: datefilter { type: date } -- Usage in Derived Table SQL: -- WHERE {% condition datefilter %} created_at {% endcondition %} ``

Parameters: The "User Input"

• Role: Does NOT generate SQL itself. It holds a user-selected value to be injected into other fields.
• SQL Context: Accessed via Liquid variables ({% parameter name %}) inside Dimensions, Measures, or Derived Tables.
• Input: User selects from a UI list or types a value.
• Example:

``lookml parameter: timeframeselector { type: unquoted allowedvalue: { value: "month" } allowedvalue: { value: "year" } } dimension: dynamicdate { sql: DATETRUNC({% parameter timeframeselector %}, ${created_raw}) ;; } ``

Dimension Groups: The "Time Generator"

• Role: Defines the source timestamp or date column. Looker then generates multiple dimension fields based on the timeframes list.
• SQL Context: Casts and truncates the source column for each timeframe.
• Input: Must be a standardized timestamp or date expression.
• Example:

``lookml dimensiongroup: created { type: time timeframes: [date, month] sql: ${TABLE}.createdat ;; } -- Generates: -- createddate -> CAST(table.createdat AS DATE) -- createdmonth -> DATETRUNC(table.created_at, MONTH) ``

3. Summary of Differences

Reference Skills

For detailed standards on specific field types, refer to:

• [Dimensions](references/dimension.md): Naming, labels, and type-specific rules.
• [Measures](references/measure.md): Aggregation types, filters, and formats.
• [Filters & Parameters](references/filter_parameter.md): Templated filters and user input.
• [Dimension Groups](references/dimension_group.md): Timeframes and intervals.
• [Value Formats](references/value_format.md): Named and custom currency/number formats.