Text output files should use .txt file extension about ocrd_fileformat

[stweil]

@kba, how should this issue be addressed? Do we require users to set the correct extension by using the ext parameter?

The ext parameter default is set in the JSON file. It is currently .xml, and I am afraid that automatic setting of the right parameter value would not fit to this default value. What about changing the default to an empty string and using that as an indicator to set the file extension based on the MIME type? Do we have other examples of OCR-D processors which determine a parameter value at runtime if no explicit value was given?

[kba]

@kba, how should this issue be addressed? Do we require users to set the correct extension by using the ext parameter?

The ext parameter default is set in the JSON file. It is currently .xml, and I am afraid that automatic setting of the right parameter value would not fit to this default value. What about changing the default to an empty string and using that as an indicator to set the file extension based on the MIME type?

That's a good idea.

Do we have other examples of OCR-D processors which determine a parameter value at runtime if no explicit value was given?

Yes, e.g. gpu_id in ocrd-anybaseocr-dewarp, which defaults to -1, which means to not use GPU at all. Also we do have the MIME_TO_EXT mapping in ocrd_utils and ocrd bashlib constants, so this wouldn't be hard to implement.

[kba]

Fixed in v0.1.0.