Redis can encode up to 52-bits of precision (the matissa of a double precision floating point) in the score when translating a sort key byte array into a score for the Redis Scored Set. When range queries are explicitly relying on the sort key range to exactly filter results without any fine-grained filtering (such as the case for spatial, temporal, spatio-temporal queries) there is only so much precision available in just the score. To fix this we will have to store bytes of higher precision and use direct comparisons.
Redis can encode up to 52-bits of precision (the matissa of a double precision floating point) in the score when translating a sort key byte array into a score for the Redis Scored Set. When range queries are explicitly relying on the sort key range to exactly filter results without any fine-grained filtering (such as the case for spatial, temporal, spatio-temporal queries) there is only so much precision available in just the score. To fix this we will have to store bytes of higher precision and use direct comparisons.