Yes - parquet tooling is non-existent compared to CSV particularly on the command line. And the cross-language/platform support is a mess - good luck reading P

Yes - parquet tooling is non-existent compared to CSV particularly on the command line. And the cross-language/platform support is a mess - good luck reading Pandas generated parquet on .NET or in a (non-spark) JVM environment. There are many reasons why CSV is flawed for the purposes of storing tabular data (e.g. loss of column type information) but the alternatives are just so unergonomic that CSV remains a viable choice in many situations.