Github user pvillard31 commented on the issue:
https://github.com/apache/nifi/pull/2966
Hey @markap14 - thanks for the review. You're right, it's much better to
address it when creating the Avro schema. That would prevent this error to
happen again if one day we add a Reader for a format that would accept non Avro
compatible field names.
Just a quick remark. Initially in ``AvroTypeUtil`` I just wanted to change
L123 from:
````java
final Field field = new Field(recordField.getFieldName(), schema, null,
recordField.getDefaultValue());
````
to
````java
final Field field = new
Field(normalizeNameForAvro(recordField.getFieldName()), schema, null,
recordField.getDefaultValue());
````
But not sure changing the name just here is really a good idea and if it's
not going to break some things in other places. In the end, I changed all the
occurrences of ``recordField.getFieldName()`` to normalize the name. Thoughts?
---
