Add fixed-schema format attributes to Generic output port of FeatureReader transformer

Hi,

some formats (such as the Text File reader) have fixed schemas that will never change. In the case of the Text File reader the only attribute in the schema is the text_line_data attribute.

For these fixed-schema formats, I would like to request an enhancement to the FeatureReader to have the schema automatically exposed, even when the Generic output port is configured. Currently, the schema attributes are only exposed for named output ports.

For context, I refer to the following Community question:

And to my own support question (to which James Cheng provided a very comprehensive reply):

https://support.safe.com/hc/en-us/requests/64419

I have reproduced James’ reply below, as it so clearly explains the situation:

Thank you for raising this. I understand the frustration as I share the same thoughts and can see why you'd expect the text_line_data attribute to just be available when reading a Text File through the FeatureReader.
 
The behavior you're seeing comes down to how the FeatureReader's generic output port works. This port is a catch-all that merges everything into a single stream, designed to handle any format and feature types with varying schemas. Because of this, it doesn't commit to any schema at authoring time, it doesn't make assumptions about which feature types or attributes will be present, so all attributes come through unexposed by default, even for predictable formats. By contrast, when you use a named output port (or a standalone Text File Reader), the schema is known at authoring time, so the attributes like text_line_data are automatically exposed. In essence, by using the Generic port, the trade-off is flexibility over schema awareness, it can handle anything but at the cost of not pre-committing to any specific attributes.
 
However, as you've rightly pointed out though, the Text File Reader always produces a predictable fixed schema with a single attribute, text_line_data, no matter which files/feature types are being read. So there's a reasonable argument that the FeatureReader could recognize this and auto-expose it on the Generic Port since there's no ambiguity what the output will be. Currently, the Generic port doesn't differentiate between fixed schema and variable schema formats as it applies the same "unexposed by default" behavior uniformly across the board.
 
That said, I think this is a genuinely good candidate for a product enhancement for the generic port. I'd encourage you to post this as an Idea on the FME Community, specifically requesting that the FeatureReader auto expose known attributes for fixed schema formats like Text Files on the Generic port and describing your use case. Community support helps the development team gauge interest and prioritize ideas for future releases, and we can link this ticket to the Idea to keep it on their radar.
 
In the meantime, it's worth noting for this specific format, the feature types are all fixed to text_line, so using the default One per Feature Type output port setting will also route all text file features through a single text_line named port, similarly to the Generic Port, which will expose the text_line_data attribute automatically. If your workflow requires the Generic port, then the only available options are manually setting the Attributes to Expose parameter on the FeatureReader or placing an AttributeExposer downstream.

regards,

Nic