Bugzilla

(In reply to Chris H-C :chutten from comment #1)
> I think to start we might wish to "migrate" one while still leaving the Telemetry in place until after verification. In other words:
> 1) Add a new Glean collection that mimics the EXTRACT'd collection
> 2) Verify that they report similar/identical data (modulo timing and representation differences)
> 3) Remove the mimic, properly migrate the EXTRACT'd collection
> 
> (( And all this maintaining the flow of data to Telemetry (ie, step 3 involves setting up [GIFFT](https://firefox-source-docs.mozilla.org/toolkit/components/glean/user/gifft.html)) ))

Yes, we can also go that way. fwiw I still posted the patch as is for now to proof it "works" as in it compiles and I see some data in a first test run.
Your staged approach will let us verify that in the real world first, so I'm not opposed to changing it to that.

(In reply to Chris H-C :chutten from comment #1)
> I think to start we might wish to "migrate" one while still leaving the Telemetry in place until after verification. In other words:
> 1) Add a new Glean collection that mimics the EXTRACT'd collection
> 2) Verify that they report similar/identical data (modulo timing and representation differences)
> 3) Remove the mimic, properly migrate the EXTRACT'd collection
> 
> (( And all this maintaining the flow of data to Telemetry (ie, step 3 involves setting up [GIFFT](https://firefox-source-docs.mozilla.org/toolkit/components/glean/user/gifft.html)) ))

Yes, we can also go that way. fwiw I still posted the patch as is for now to proof it "works" as in it compiles and I see some data in a first test run.
Your staged approach will let us verify that in the real world first, so I'm not opposed to changing it to that.

Re step 3: No need! `geckoview.version` and `.build_id` are legacy telemetry probes for the sole purpose of being EXTRACTed. They don't appear in legacy telemetry at all.