Dynamic Excel schema not refreshing when column names or new columns are added

As far as I know, dynamic reading and writing just works. Just doublechecked in 2025.2.1 with a simple workspace reading 2 different Excel documents with a changed column name and an added column and writing this to a resulting CSV. So dynamic converting is not an issue. Just use a FeatureReader reading both data and schema and write these to a FeatureWriter set dynamic. See attached sample.

But it is hard to do something with these changed columns as dynamically exposing unexposed attributes is not possible. This makes sense if you realise transformers downstream of the reader can only be assigned to exposed attributes, so to handle these attributes you will need them before runtime.

There are ways work with this, like the BulkAttributeRenamer regexmapping identification columns to a fixed ID column before exposing it. Or exploding the table to features using an AttributeExploder, renaming columns, rebuilding the data using an AttributeCreator to dynamically write both attribute name and attribute value. Or the SchemaScanner to generate a new schema feature before writing it to a dynamic writer.

But it all depends on what you need to do with the features after reading the data. What do you need after de features are read?

I will need to unpivot the data and join it with data from another table before writing it. So I can’t unfortunately just send it into a dynamic writer.

 

There is an additional step you need to do when adding the excel reader to get it to read dynamically by selecting sheets to read to none

Dynamically Read Multiple Excel Files – FME Support Center

​​​​​​​

Regarding your question 3, if the process expects a fixed schema the excel file is read dynamically and the schema is checked to ensure certain required fields are present, if they’re not present a message is returned to the user and no further processing occurs.

I already select sheets to read to none, still have to press “Update reader” in the navigator pane to make it read in the new/changed column names. 

 

Are you using a regular reader rather than the FeatureReader?

I see that the attribute names are automatically exposed on the reader even when set up dynamically which I think is incorrect. But the reader does read the updated data irrespective of whether you’ve updated the reader.

e.g. updating a single column name, you will find the value of the exposed attribute that no longer exists will say missing and the new column name will be present but unexposed.

Updating the reader just controls which attributes get exposed.

You cannot dynamically expose attributes