The latest version of our snowplow web package, version 0.13.0 has just been released. This release has some large changes including how we processes sessions (which now include all stray page pings), and adds support for redshift/postgres and databricks for our consent models meaning we now support all 5 major warehouses with these models (all credit to @Agnes!). You will need to make sure you’re using at least RDB loader 4.0.0 now if you’re on redshift/postgres, or want to use the consent modules (as we now require the load_tstamp
column).
Because of some of the changes we have made, you’re very likely to see changes in session counts between previous versions and this new version, so make sure to read below to find out how these changes will impact you and why we made them.
Summary
This release adds Databricks, Redshift and Postgres support for the optional Consent module. As a breaking change we are decoupling sessions from pageviews, please refer to the below breakdown on this. It also removes unnecessary domain_userid
tests which would fail due to anonymous tracking. We also fix some integration tests.
Breaking Changes
- The way we generate the sessions table has changed, breaking the dependency on the page views table and instead building directly from page view and ping events. This means we now include sessions that contain only page-pings, and all stray page pings in sessions, see our docs for more information about what this means. You do not need to make any changes to your tables or data, but you will likely see an increase in the number of sessions overall, and may see a small increase in absolute and engaged time on individual sessions (to account for those starting with a stray page ping). Any reporting off the sessions table should be interpreted accordingly, or the data re-processed (see here for how to reprocess a single model).
Prerequisite Alert
Redshift/Postgres users would need to use RDB Loader v4.0.0
and above as the web model for them relies on the load_tstamp
field from now on. The same dependency applies for the Consent module users regardles of warehouse as that field is utilised for the incremental logic.
Features
- Fixed sessions being dependant on snowplow_web_user_mapping even when session stitching was disabled (Close #155)
- Added alias argument to the
filter_bots
macro for more flexibility - Migrated engaged time to use device timestamps instead of derived to provide more accurate calculation when there is a delay between events being sent and the collector receiving them
- Added
get_optional_fields
macro to reduce duplication of calls to contexts across models - Added missing adapter support for Consent module (Close #148)
Under the hood
- Tidied up some macro inheritance
- Improved macro documentation
- Improvements to integration test
Upgrading
To upgrade bump the snowplow-web version in your packages.yml
file, taking account of the breaking changes above.