1934939 - [SPIKE] Allow for action button ordering to be configurable

Status: NEW

(Firefox :: Messaging System, task, P2)

Description

[Negin Sauermann]

For feature callout messages, we commonly have designs where the secondary button (split dismiss) is on the right and the primary button is on the left. To accommodate these designs, we use the additional_button styled as a secondary button, and the secondary_button styled as a primary button because of how they are ordered in the code:
https://searchfox.org/mozilla-central/source/browser/components/aboutwelcome/content-src/components/MultiStageProtonScreen.jsx#152

It may make more sense to refactor our button components to handle ordering more intuitively through some sort of property and perhaps further abstracting our buttons.

This also has an impact on data monitoring and automating dashboard creation:

... we have no primary click telemetry for the message. This means that in Skylight (as currently coded) it just so happens to work that the "secondary" in the JSON is displayed on the graph as the primary, but as far as I know, that's only because of a random(?) implementation quirk of Looker. Which is to say it's quite fragile.
Before knowing what you just said, I assumed that if there was more than one button, there would be a primary and a secondary as known to the telemetry, but that's not the case.

Comment 1

[Negin Sauermann]

I know we've discussed this in the past, but I wonder what your thoughts are on this Shane and whether this use case feels more specific to Feature Callout than our other surfaces? Or does it make sense to tackle this for all surfaces? I can see this getting a bit hairy with about:welcome.

Comment 2

[Dan Mosedale (:dmosedale, :dmose)]

In addition to the stuff quoted in comment zero,

• the meaning of the word "primary" in the Feature Callout JSON was confusing (the implication being that if there were one or more buttons, the first one would be the primary button), which made it non-obvious why there wasn't such a button (ie if that was bug)
• I ended up spending about an hour playing with Looker and STMO to get enough context to understand whether Skylight was working correctly

Comment 3

[Negin Sauermann]

Not being actively worked on, removing myself in case someone else has capacity to grab this.