There are a few problems with the current schema UI.
- It is not immediately apparent which tables may have columns unselected. A user has to manually go through each table and check for a partial selection icon for columns. This is a time consuming issue for datasets which may contain hundreds of tables. Suggested fix is to have the partial selection icon trickle up the schema tree so a user is immediately aware that a dataset is only a partial sync and which tables will not be syncing all columns.
- There should be a method to select all recursively from the dataset or table selection so that all columns are selected. Current behavior is that reselecting a table with unselected columns doesn't reselect all the columns in that table. So a user must go through individually to ensure columns are selected. Again, this is an issue for datasets with hundreds of tables especially in combination with issue 1.