It crossed my mind that we might be able to produce sample code for doing a sequence test, then hand off to the banner team to play with. However, they might not have enough time to make this a worthwhile approach.

Still, it seems like the rules for conditionals and banners to display should be editable by the banner team, maybe a hook supplied by the banner? Gah.

Initial consensus in today's tech priorities meeting was that we should avoid anything involving buckets even for the trial implementation.

Here's what I have so far:

• This should be implemented as a mixin (previously discussed as a likely option).
• Some tweaks will be necessary to the banner choosing code to allow mixins control over banner choice. Maybe we should create a new hook for this mixin.
• More tweaks will also probably be needed to the mixin parameter system, including the UI... Unless we could make do with a fixed number of steps in the sequence. I guess this might be OK for an initial MVP, but it seems less than ideal!
• Finally, the output of Special:BannerAllocation should also be tweaked to show multiple banners for each bucket.

I have the feeling that the changes needed to CN to support the fallback feature are more basic and logically go first. However, before coding on either, it might be helpful to have a mock-up of how Special:BannerAllocation might look once both fallback and banner sequence features are implemented.

Thanks!!!

@AndyRussG
That sounds good--I'm curious what the sequences will look like. We're talking about testing two sequences against each other, I understand. So the admin will have to configure two lists of banner names? There will be validation or autocomplete help to make sure the banner names are correct?

Spec:
https://docs.google.com/document/d/1YlNwpFZ4gPyO1BYxYRE81F6n0KHeB0itfwVnLcbB1EA/edit

Here are the notes we have for an MVP. I don't know if we should leave this task open until we finalize details of a more usable version...

This is fine to resolve. Thanks for working through this!