1957476 - Move extended attribution and distribution object metrics `glean.attribution|distribution.ext` to `client_info` alongside the core distribution and attribution fields

Status: RESOLVED FIXED

(Data Platform and Tools :: General, task)

Description

[Chris H-C :chutten]

In bug 1955428 we are implementing the Glean SDK attribution and distribution instrumentation that was designed in this doc in bug 1930762. One of the pieces of this design is a pair of object metrics, glean.attribution.ext and glean.distribution.ext which, rather than ending up in metrics.objects., are intended to live alongside their core attribution|distribution fields in client_info.{attribution|distribution}. as client_info.attribution.ext and client_info.distribution.ext.

The plan in the design doc was to use column replacement akin to the client_info normalization already being done in mozilla/bigquery-etl's sql_generators/stable_views/__init__.py. I made an attempt at it in https://github.com/mozilla/bigquery-etl/pull/7252 which ended up not working due to SQL's REPLACE needing special care for nested columns.

This bug is about making that replacement happen, or to otherwise satisfy the design's goal of having all the core and extended attribution and distribution data all together in datasets for ease of comprehension and use.

Comment 1

[BMO Github Automation]

Attachment: [mozilla/bigquery-etl] Bug 1957476 - Add attr/dist metrics to client_info in stable views (#7293)

Comment 2

[Ben Wu [:benwu]]

client_info.attribution.ext and client_info.distribution.ext have been added to all the stable table views. I don't see the metrics in any of the apps yet so those columns are currently just null jsons and I can't verify it's completely working but it should be fine