BUG qsv diff produces different results for the same command

[datatraveller1]

This is an interesting issue. Have you noticed that successive invocations of the same command with qsv diff give different results? The results are usually correct, but sometimes wrong.

qsv diff --key=art_no a.csv b.csv -o diff.csv

If you call this command twice or more, the file diff.csv often (but not always) contains different content.

I have not yet succeeded in creating a small test file without confidential data, but perhaps you can already do something with this information.

[jqnatividad]

Thanks for the heads up @datatraveller1 . I know it uses parallelized hashing, which may account for this non-deterministic behavior.

Copying in @janriemer.

He contributed the diff command, which takes advantage of his csv-diff crate...

Hopefully, he can quickly identify the reason why diff sometimes produces different results...