fix: don't track exposure to fallbacks and add tests #9
Conversation
|
Why shouldn't we track exposure to fallback variants? |
I posted about this in #skylab-internal My guess is that a customer will never use an exposure to a fallback metric as the exposure metric in an experiment. Therefore it is not useful to send. For debugging, tracking exposures to fallbacks would be useful. Nirmal brought up that we could send a different metric in this case (e.g. |
| return { value: undefined }; | ||
| } | ||
| const sourceVariant = this.sourceVariants()[key]; | ||
| if (sourceVariant?.value) { |
There was a problem hiding this comment.
Why not if (sourceVariant) {?
I had to check but we do currently require a value so this existing code is correct.
There was a problem hiding this comment.
Technically, sourceVariant could be non-null with a null value, although this is definitely not the regular case and I'd doubt it'd ever happen. That being said, a null value means an invalid variant, which we do not want to track.
Summary
Checklist