> And because often also the error handling is incomplete it might be difficult to know that the pipeline is producing garbage output. The thing is that from t

> And because often also the error handling is incomplete it might be difficult to know that the pipeline is producing garbage output. The thing is that from the point of view of each individual tool, there isn't even an error! You ask awk to split on commas and give you the second field, it splits on commas and gives you the second field. It opened the file no problem, it didn't run out of memory writing its output to stdout... There were no errors to speak of from what it could see. The "error" is an error in the overall logic of the code.