We have many SSIS packages that sort incoming data, and merge-join to other tables from other systems.
Some of these packages may have 10 or so joins (each preceded by a sort on the join fields).
In the case where one or more new fields is added to the source, we then need to tick the pass-through box for each new field, on every sort step, and include the new field on every join output. In the case where a column is removed, we need to select every transition line and delete the old, unbound input.
This is a fiddly, error prone, RSI inducing operation that is really just busy-work.
Does anyone know a better way? Is there a way to automatically pass every input field as an output field, on every transform step?
Or to convince the IDE to go through and remove an old unbound input from every transition? (In 2008r2 and possibly later - deleting or remapping a column input would flow through... not any more it seems?)
It seems to me that the better general case would be to have fields pass-through from step input to output as the default, and explicitly "block" fields that are not needed later by manually un-checking them. And even then - since the run-time warns of unused fields, couldn't we have an option to detect all unused fields, and optionally block these after their last use, at design time?