a current weakness in sybil is that if a column has a type beforehand (int or string), samples that come in with a different type will have that column discarded (but the rest of the sample will be retained).
this means that once a column is created, surgery is necessary to change its type. its much easier just to add a new column with the new type.
taking care of this task will make it so that this problem no longer exists, whatever solution that is.
a current weakness in sybil is that if a column has a type beforehand (int or string), samples that come in with a different type will have that column discarded (but the rest of the sample will be retained).
this means that once a column is created, surgery is necessary to change its type. its much easier just to add a new column with the new type.
taking care of this task will make it so that this problem no longer exists, whatever solution that is.