This page refers to the
typeparameter that is part of a
typecan also be used as part of a dimension, described here
typecan also be used as part of a measure, described here
typecan also be used as part of a dashboard filter, described here
typecan also be used as part of a dashboard element, described here
type is a child of
join is a child of
joins is a child of
a join type:
type enables you to describe the type of
join that you want to perform. The four possible values for type are:
left_outer (the default value)
The most common join type in Looker is
left_outer (the reason for its frequent use is described below). You may know it from SQL as a
LEFT JOIN. When using this
type, all rows from the explore are included and data from the joined view is brought in, if available.
full_outer join type — which you may know from SQL as a
FULL OUTER JOIN — includes all rows from the explore and joined view, even if there is no corresponding data from the other view. Please note that full outer joins are not supported in MySQL.
inner join type — which you may know from SQL as a
INNER JOIN — only includes rows that have matching records in both the explore and joined view.
cross join type — which you may know from SQL as a
CROSS JOIN — is rarely used. It is sometimes helpful for zero-filling or matrix generation. It creates a row for every combination of the explore and joined view. The concept is difficult to show on a Venn diagram, and may be better understood by looking at the table example:
Previous to release 4.0 one common use case for cross joins was to join a date table into your data so that a row was shown for every date, even if there were no results. Looker’s dimension-fill functionality can now typically fill that need.
LEFT JOIN to add dna data to your people data, if dna information is available:
FULL OUTER JOIN to add user data to your order data, even if the user hasn’t placed an order:
INNER JOIN to add user data to your event data, and limit the results to just the events that were generated by a user:
CROSS JOIN to generate all possible combinations of owner and property:
Things to Know
left_outer Join Types Are Usually Preferred
When learning and using SQL, many people focus primarily on
JOIN, which is technically known as an “inner join”. For this reason, one might assume that
type: inner_join should be used in the majority of cases. However, Looker typically works best with
type: left_outer, which is like using a
LEFT JOIN in SQL.
The reason for this is that the view associated with an explore (as opposed to the views joined into an explore) is usually the “primary” data. For example, you might have an explore based on event information. Some of the events might be generated by your system and other events might be generated by your users. If you join user into event using a regular
JOIN, you will only be able to see events that were created by a user, and you’ll lose visibility to system events. By using a
LEFT JOIN instead, you’ll be able to see all the events, whether or not there is a user associated with them.
You May Want To Use
always_join If Restricting Rows Is Important
When Looker generates SQL for a query, it attempts to create the cleanest SQL possible, and will only use the joins that are necessary for the fields a user selects. When you use
type to define something other than a
LEFT JOIN, you might always want a certain join (or set of joins) to be part of the query, to make sure the resulting rows include your join restrictions. You can achieve this by using the