If first 5 values of columns are the same, the seed is the same. Should it be different? #916
Description
Problem Description
RDT is supposed to be setting the random seed based on the first 5 values in the column. For a lot of cases, this is sufficiently allows users to:
(a) transform/reverse transform in a reproducible way while also
(b) creating different data for each different column
However, this would break down if you have 2 different columns and the first 5 values of those columns are exactly the same. In such a case, the random seed for those columns would be exactly the same, so whatever randomness they have will be in-sync.
This is a general issue that we can use to track the problem. If it ends up affecting a lot of users, we may want to consider finding a different way to set the seed.
Additional context
This issue assumes that #906 has already been fixed. (#906 identifies an bug where the first five column names are being used rather than the first 5 data values.)