Currently, when using AttributePivoter the resulting features have the "pivoted" attributes unexposed. This behavior does not replicate what some users would expect from a "pivot." For example, when creating a Pivot Table in Excel or using popular R packages reshape or tidyr, one expects the resulting pivoted table to contain exactly the rows, columns, and cell values specified. The resulting table of exposed attributes in FME does not look like that.
Therefore, users currently must add an AttributeExposer after the AttributePivoter. That's well and good, but why not add an option to expose attributes in the AttributePivoter to avoid this step? This section in the transformer could mirror the "Attributes to expose" parameter in the XMLFlattener or JSONFlattener.
Even better would be allowing these parameters to get their value from an attribute or through an Import feature such as is available in the AttributeExposer. See this related idea.
While we are at it, do any other transformers need an "Attributes to expose" parameter that currently don't have one? I see a few existing ideas:
