ping @oreoshake

Apologies, this fell off my radar. I'd have to dig around a little more to make sure there wouldn't be side effects to these decisions and I remember having a little trouble keeping the edge cases in mind here. But at a quick glance, this would be my first guess at desired behavior.

CSP Config Report Only Config Apply Config Result

PRESENT OPT_OUT :report_only ?
OPT_OUT PRESENT :enforced ?
OPT_OUT OPT_OUT :both ?
OPT_OUT OPT_OUT :enforced ?
OPT_OUT OPT_OUT :report_only ?

Raise an error? These entries all seem contradictory.

PRESENT OPT_OUT :both ?
OPT_OUT PRESENT :both ?

Raise an error? While these are partial contradictions, it could lead to surprising behavior. I would also accept a noop. Something about :both being partially inaccurate seems to be worse than e.g. OPT_OUT OPT_OUT :enforced which doesn't seem to make a lot of sense in my head.

OPT_OUT OPT_OUT nil ?

Noop? No lies were told here. Because the default value is nil, raising an error in this case would require someone to opt out of CSP three times 😄

Again, apologies if I'm way off here but have we considered having two separate values that would both need to be set rather than this report_only/enforced/both situation? I think that would give us an a strong case for raising errors for all of these question marks.

Thanks for the issue, PR, and followup @pcasaretto. This is back on my radar.

For transparency: I've left the company I was working for and this is no longer a priority for me.
I'm closing it for now.