/shrug I like CSVs. Or, rather, TSVs. They're simple, structured and interoperable. It's the first format I reach for when building small pipelines. The fault

/shrug I like CSVs. Or, rather, TSVs. They're simple, structured and interoperable. It's the first format I reach for when building small pipelines. The fault is not with the format, but with the developers that choose to use it after they've outgrown it. For anything more complex, choose SQLite. For everything else, choose Postgres. Use the right tool for the job. There's likely much worse things happening in these cases than just broken CSV parsing. Since these issues got deployed to production, lack of testing, for one.