Looker 6.8 Release Notes

  • 5 March 2019
  • 3 replies

Userlevel 5

Release Rollout Begins: March 10, 2019

Release Final & Download Available: March 20, 2019

Release Highlights

In addition to general tweaks and enhancements, this release comes with new and improved features in the following categories. Read on for more detail.

Preparing for Release

Please take notice of items marked with a ⚡ as they indicate changes to existing functionality and may require your attention. For more information see the Legacy Feature Updates and Features by Section below.

Experimental and Beta Features

The following new and improved experimental and beta features are marked with a ⚗:

Notable Features

Localized visualizations

Dashboards and visualizations can now be displayed in French, German, and Japanese, giving your content worldwide reach! Embedded content will pick up a user’s locale. Filters, Looker menu items, etc. will not be translated.

Import private repositories

The experimental Project Import Labs feature now supports references to private external repositories. Use it to reference and extend projects from elsewhere in your organization or code blocks anywhere you can access them!

Features by Section

LookML and Development

  • Remote project import for private repositories. The experimental Project Import feature now supports references to and extensions of private git repositories. Learn more.

  • Explicit LookML field type references. Introduced a new syntax for referencing LookML field types other than the default, for example you can now reference a datetime version of created_date with ${created_date::datetime}. Learn more.

    • ⚡ As part of this change, field references to list measures are not supported and will produce errors.

  • The Content Validator is now a fully supported feature.

  • “Don’t Revalidate Unchanged LookML” has also graduated from Labs. Learn more.

  • Hyphens are now allowed in string file names.

  • Improved IDE performance for very large LookML projects.

Dashboards, Visualizations, and Explore

  • Localized dashboards and visualizations. Visualizations and dashboards now support French, German, and Japanese. Application menu items, filters, etc. do not yet support translations. Learn more.

  • ⚡ ⚗ To be consistent with how dashboards are handled, the beta API 3.1 no longer includes soft-deleted Looks in all_looks() calls. The call is not changed in API 3.0.

General Tweaks and Bug Fixes

  • Dialects

    • Fixed an issue where Druid connections on clustered instances would produce Cannot run simple select querysimple SELECT query tests Java::OrgApacheCalciteAvatica::AvaticaClientRuntimeException: Remote driver error: ISE: Connection8b4d9fb4-2383-4aac-92e0-e2b5231aa0e1 already open errors.

  • LookML and Development

    • Fixed an issue where Looker did not escape commas in parameter field values.

    • Fixed an issue that produced “should be localized but has not been” IDE warnings for LookML dashboards when string files are present.

    • Fixed an issue where the LookML Validator failed to run when remote_dependencies did not have url and ref parameters.

    • Informational messages related to localization of LookML dashboards are now grouped in the IDE as “LookML Info Messages”. This addresses a previous issue that treated them as warnings even when permissive localization was enabled.

  • Content Management and Discoverability

    • Fixed an issue that prevented enabling a closed system in the presence of orphaned Looks and dashboards.

    • Fixed an issue where the Explore picker drop down could be cut off on small windows.

    • Fixed a closed system issue where non-admin users could lose access to a Space when adding users to it.

    • Addressed issues that could cause broken Look and dashboard thumbnails to appear in Spaces.

  • Scheduling and Downloading

    • Fixed an issue that caused jobs to fail when cluster nodes listened on different ports.

  • Embedded Analytics

    • Fixed an issue where menu items produced 404 errors in the JavaScrip Console.

3 replies

Userlevel 3

Regarding Explicit LookML field type references: Does this mean that we will be able to prevent Looker from automatically casting?

For example, say I’m joining two views on dimensions of type: date. But the underlying data is of variable type, my database handles the conversion to date slowly, and I’d want to join on strings instead, with SQL left(field1::varchar,10)=left(field2::varchar,10). Will the Looker syntax left(${field1::varchar},10)=left(${field2::varchar},10) accomplish that? Or will that Looker syntax instead generate the SQL left(field1:📅:varchar,10)=left(field2:📅:varchar,10), and thus not help in terms of speed at all?

Userlevel 7
Badge +1

At least in BigQuery, it looks like it’s not doing that. It’s converting to string with a FORMAT_TIMESTAMP, but is still doing the inner date cast.

::string SQL

(FORMAT_TIMESTAMP('%Y-%m-%d', TIMESTAMP_TRUNC(CAST('2019-01-05'  AS TIMESTAMP), DAY)))  AS ascent_datevariabletest

vs no explicit field type reference SQL:

(CAST('2019-01-05'  AS DATE))  AS ascent_datevariabletest
Userlevel 3

Thanks for the info, @izzy. That’s a shame. But of course we can always define another dimension as the same column cast to string. I was merely hoping that this would obviate the need to do so.