Rationalize KVP controls

[bretg]

Type of issue

breaking change

Description

Controlling PBJS ad server KVPs (aka "targeting") is quite complex. There are 12 different options publishers can define to refine what's sent. There are often edge-cases discovered where the interaction of several of these options don't work properly. This isn't surprising because we've never sat down to rationalize the whole mess of control options and how they're supposed to work together. Instead, people have just organically added yet another control randomly into the code to meet their specific use case.

The range of controls is documented at https://docs.prebid.org/features/adServerKvps.html

I've drafted a Prebid.js Ad Server KVP Control doc with a first cut at a coherent description of how the 12 controls work together. I'm certain that the current code doesn't work this way, and suspect that there will be some interesting debates about semantics.

I'd like to discuss as a community:

1. Do people agree that the fringes of KVP control is a somewhat mysterious and idiosyncratic ritual?
2. Are we willing to take a breaking change in this part of the system conform to some agreed-upon algorithm?

[bretg]

Discussed in committee last week. We agreed that spending time in this area would be worthwhile, but it would more of a project for PBJS 8.0.

[dgirardi]

Here's some related bug reports:

#6734
#8052

Their fix requires either modifying the effect of some of the current controls (which is a breaking change), or introducing more controls. The first route is obviously preferable, but are we OK with pushing it forward to 8.0? Should we consider a narrower redefinition of the "broken" controls for 7.0?

[patmmccann]

Work is not currently planned; people interested in this work are invited to express support and we may plan the work.