Composite Samplers Proposal needs clarifications

[PeterF778]

What are you trying to achieve?

The Composite Samplers Proposal attempts to bridge Consistent Probability Samplers with legacy samplers by looking at the parent's sampled flag when no rejection threshold is provided for the parent span. This is expected to correctly handle (albeit with no Consistent Probability Sampling benefits) situations when parent spans are sampled by a non-consistent samplers and the user intention is to follow the parent's sampling decision.

However, the paren't sampled flag may itself depend on the trace randomness value, which can lead to sampling decision not aligned with the user intentions.

The proposal needs to clarify that when making a non-consistent sampling decision, the original trace randomness must not be used.

_{Tip: React with 👍 to help prioritize this issue. Please use comments to provide useful context, avoiding +1 or me too, to help us triage it. Learn more here.}

[PeterF778]

Please assign this issue to me.

[danielgblanco]

@PeterF778 I've applied the sig-issue label, and added the task to the Sampling SIG board, to reflect that this is something that the Sampling SIG will be tackling. Let me know if that is not correct. I don't think this issue in the spec needs to assign, as long it's on your board and labelled appropriately it's okay.