Hey team, I think I may have found a bug introduced in the latest update. So I’m using an anonymous_id for the experiment configuration, and then I have a user_id only metric that make use of the identifier join table defined in the source. This stitching works perfectly. However, when I add an activation metric (also user_id only) it no longer joins the

__identities0

CTE to the user_id only metric I mentioned above. I’m pretty sure this was not the case before I upgraded to the new version. What do you guys think?

hi Malcolm, thanks for reporting this

Just pushed a fix for this. A new docker build should be available in about 10 minutes. It was choosing

user_id

as the base id type for the query since it was used the most times (both for the goal and activation metric). Now it always chooses the base id type from the experiment assignment data.

Thanks, I just upgraded and everything works again!