Annoying File Reader behaviour

KNIME Analytics Platform
1

Hi,

Could someone explain why is it useful if the Column delimiter selection change also changes in some cases the read row IDs property in File Reader node? (Steps to reproduce: open configure, select a file (from history), read column headers, change the Column delimiter to <tab>. There are other ways, when the uncheck of read row IDs is unchecking the read column headers checkbox too.)
I am a bit confused…
Best regards, gabor

2

The “read rowIDs” setting should be fix after you set/unset the check mark there. Until that, the filereader guesses the correct value for this (it looks at the first column in the file and checks if the values have increasing indices) and that could certainly depend on the delimiter (because the values in the first row could change with a different delimiter).
Same thing for the read column headers. It is free (and set by the file reader) until you manually set it.
That’s at least the intended behaviour. I hope it is consistent with what you see.

3

Just to be sure. This is intended behaviour:
I have a file with tab separated header, that is the only line (no spaces).

  • Open Configure
  • Select that file
  • No checked checkbox, except ignore spaces and tabs (Remember, it is tab separated.) *
  • Select <tab> column delimiter
  • checked read row IDs, read coloumn headers and ignore spaces and tabs, the preview is consistent with this behaviour. *
  • uncheck read row IDs
  • unchecked read row IDs and read coloumn headers; the ignore spaces and tabs is still checked *
  • Now I have to check the read coloumn headers