I guess solving #81 introduce a more serious bug when using split type existence with signed.hash = TRUE. The sign of the hashed feature might not be consistent. (test case required!)
Due to the design of the internal structure, I am wondering if we will degenerate to wontfix #81. In my opinion, the #81 should not affect the accuracy of the model, so it is less important. But this bug might affect the accuracy.
I guess solving #81 introduce a more serious bug when using split type
existencewithsigned.hash = TRUE. The sign of the hashed feature might not be consistent. (test case required!)Due to the design of the internal structure, I am wondering if we will degenerate to
wontfix#81. In my opinion, the #81 should not affect the accuracy of the model, so it is less important. But this bug might affect the accuracy.