I have recently been working with origins what ingest delimited files with no header. As such, the fields come into the pipeline with names that are derived from a sequential number list. Since such field names are useless when dumping records into a database, a rename operation must be performed. I have already stated in another thread that the renaming operation is expensive, and it would be ideal to avoid if possible. For this situation it seems ideal to me to provide a field name list via the origin that could override the numerical field names. In this fashion the correct field names can be assigned to the records as they are being read from the file and avoid the need for a renamer stage.

What say others on this? Are there maybe other ways to solve this? Perhaps the resolution is to make the renamer more efficient.