The SchemaScanner is a very useful tool in dynamic workflows. However if dynamically creating an output dataset from an attribute [ex. FeatureWriter Dataset:@Value(_outputfilename)] the <Schema> feature from the SchemaScanner does not contain this attribute and the writer fails because it can't find the schema.
No definition found for input feature type '...
In most transformers, attributes included in the group-by parameter are propagated onto the output feature. If this were the case with the schemaScanner, then the required attributes can be included in the group-by and the dynamic writer would work as expected. No additional transformer required.
As it is, a FeatureMerger or equivalent is required to attach the desired attribute(s) onto the schema feature(s), which adds extra overhead to the transformation, and can cause issues with the schema features arriving first. To account for that a featureHolder needs to be placed on the Output data stream