As part of the NaN conversion audit of replication, we found that passing in w: NaN as an argument for setting write concern, the NaN is evaluated as 0. This is because we use safeNumberLong() to extract the "w" value, and this function evaluates NaN as 0.
This issue exists all the way back to 4.4.
- is related to
-
SERVER-68309 Investigate for unsafe narrowing conversions
- Closed