Looker 4.2 Release Notes

  • 17 November 2016
  • 6 replies

Userlevel 5

Release Rollout Begins: November 27, 2016

Release Final & Download Available: December 8, 2016

Preparing for Release

Please take notice of items marked with a as they indicate changes to existing functionality and may require your attention.

Notable Features

Content Management and Discoverability

You and your coworkers are probably whipping up some great content in Looker, but sorting through all that great stuff wasn’t always trivial. Introducing a new system for content discovery! Search is much zippier than the one in prior releases and is broader-reaching. A Favorites system allows you to keep tabs on the Looks and drag-and-drop Dashboards that matters most to you regardless of where they live, provided you’ve been given access of course. Last but not least, the Top Content section surfaces what you and your coworkers have been using the most. Explore on!

Schedule Full Result Sets to Amazon S3

Shirk the constraints of email and send query results to S3. Introducing the ability to send full result sets to S3, provided a query meets the Looker’s standard result streaming requirements. Bypass browser, memory, and email limitations by scheduling all results to S3 or use the process to store snapshots, periodic reports. This functionality requires the send_to_s3 permission.

Legacy Feature Updates

The following feature have been deprecated and will only be available by enabling the associated legacy feature:

Details on how to migrate off legacy features can be found here.

Features by Section

Content Management and Discoverability

  • Favorites. A system for selecting and quickly navigating to the Looks & Dashboards you care about most. Clicking the ❤ icon will put a link to the content piece in a list that’s easy to access straight from Spaces. Learn more.

  • Top Content. A list of the most popular Looks & Dashboards on your Looker instance. This section is a great way to learn about new reports you haven’t taken a look at yet. Learn more.

  • Global Search. A search function in the navigation bar that searches across Looks, Dashboards, LookML Dashboards, LookML files, Explores, your Favorites, Top Content, and pages like Browse and your Account Settings. Learn more.

Scheduling & Downloading

  • Added the ability to Schedule Unlimited Results to Amazon S3. The schedule modal now has an S3 option, to which users may schedule Look results. To see this option in the schedule modal, the user needs to have the send_to_s3 permission. An “all results” option is available in cases where query results may be streamed. Learn more about scheduling to S3. Learn more about streaming in Looker.


Streamable In-database Pivots. Implemented streamable pivots for the majority of SQL dialects for which Looker supports query streaming (exceptions in the second link below). Baring other restrictions, query results are no longer limited to 30,000 cells and users may download results without fearing undue harm to their Looker instance. Learn more about streaming pivots. Learn more about streaming in Looker.


  • Added the ability to focus on a single line in a chart when hovering. Learn more.

  • Implemented logic to reduce the chances of value label collisions.

LookML and Derived Tables

  • Referencing an aggregate measure in another aggregate measure now produces a LookML Validator error. This behavior is not valid SQL, but the LookML validator had previously allowed it to pass and fail at query time. Learn more.

  • Introduced the sql_always_having parameter, which conceptually behaves like the sql_always where parameter but instead affects the HAVING clause of a query. The parameter can be used to limit the granularity of results an end user may see. Learn more.

  • Introduced the allow_fill dimension parameter, which prevents the filling when relevant dimensions are included in a query. Learn more about dimension filling.

  • Scheduled Persistent Derived Table Maintenance. Admins can enter a cron string to govern when maintenance operations such as the trigger checker and derived table table reaper run. Learn more.

Looker Design

  • Renamed the Look Validator the Content Validator. When enabled, it’s still found under the Develop tab and functionality remains the same.

  • IDE improvements.

  • Added syntax highlighting to SQL blocks and LookML objects referenced in the sql parameter.

  • Improved readability by deemphasizing LookML punctuation such as { and ;;.

  • Surfaced IDE keyboard shortcuts.

  • A personal Space will now appear at the top of the Space list in the Spaces sidebar when it is set as a default space.

General Tweaks & Bug Fixes

  • Introduced the Default query result persistence is 5 minutes legacy feature. New instances and those with the legacy feature unchecked will default to a 60 minute query result cache. It is still possible to adjust query cache timing with model- and explore-level persist_for parameters. Learn more.

  • Dimension Fill improvements

  • No longer apply fill when filters are applied to measures.

  • Dimension filling is disabled when a non-default order_by_field is used.

  • Filtering on an enumeration dimension disables filling.

  • In the event that dimension filling is not possible, queries and downloads will warn and not apply filling rather than fail.

  • i__looker updates

  • Usage panel dashboard tables no longer scroll. Users may still access more data by clicking Explore from here on each element.

  • Added tracking for dashboard usage to the Usage panel and the history explore.

  • LookML, Converter, and IDE fixes

  • Fixed an issue where links didn’t work on measures.

  • Addressed reference checking for value_format_name and map_layer_name.

  • Fixed an issue where new LookML did not recognize day_of_year.

  • The LookML conversion now automatically translates date_date to date.

  • The LookML converter will no longer fail when there are ambiguous names.

  • Fixed an issue related to pasting into the IDE.

  • The IDE no longer autosuggests immediately upon return.

  • Visualization fixes.

  • Fixed an issue related to drilling from pie charts.

  • Fixed an issue related to axis label rotation on bar charts.

  • Fixed several trend line related issues.

  • Rendering fixes.

  • Fixed an issue where text elements could break LookML Dashboard downloads.

  • Fixed an issue where automagic heatmaps could not plot data.

  • Dialect-specific fixes.

  • Fixed an issue with MSSQL related to using a relative date filter with a non-US server default date format.

  • Fixed an issue related to using Custom Filters in Vertica.

  • Fixed an issue related to Symmetric Aggregates on BigQuery Standard SQL.

  • Fixed an issue where LookML Dashboards could not be copied into a Space.

  • Fixed an issue where the LookML Validator errored if the raw timeframe is used in a set of filters.

  • Fixed an issue where the save modal would not load for certain users.

  • Fixed an issue where the chat window duplicated on the Look page and the Edit modal.

6 replies

Hello, how do you determine “Top Content”? There must be a time range for that. Is it last month or last X days? Is it based on number of times used or total minutes spent on the content or number of unique people used?

Userlevel 5

We started simple, at the moment, there isn’t any timeframe for the content. It’s a straight count of page loads on a piece of content. For example, loading one dashboard with 12 elements counts as one hit rather than a dozen.

So top content on the Looks part will only be direct access to the Looks? If you use a dashboard that contains a Look, that won’t count as a hit for that Look, right?

Userlevel 5

Yeah, in that example credit would go to the dashboard.

I’m a bit concerned that the move to 60-minute default persist_for means that Looker is becoming less suitable for some use cases. E.g., we maintain a dashboard of tasks in progress; it’s based on some complicated SQL that we love having developed in Looker (and maintained as PDTs, etc). Our online systems have recently been updated to use the Looker API to query a Look to help make decisions about new task assignments. At 5 minutes, these already feel a bit out-of-date. At 60, they’re useless… and represent an engineer week or two we’ll have to re-invest if Looker is going to be even more focused on slow data than previously; we’ll also have to make a more strategic decision about our fast data solutions sooner than I’d hoped to be forced into.

Hi Rick! Those are valuable points to bring up. Given the close relationship between both our companies, I want to make sure I highlight our vision here.

Our adding of a 60-minute default does not represent an abandonment of real-time querying. For most of the use cases in our customer base, 60-minutes will result in less DB hits and an overall greater experience.

Yes, we are optimizing for most use cases, but our commitment to supporting real-time querying remains.

With that said, I’ll be reaching out to understand your points even further privately.