Hacker News Reader

If we’re imagining a parallel universe where things make sense, let’s get rid of CSV files altogether and use the ASCII field separator characters. https://en.wikipedia.org/wiki/C0_and_C1_control_codes#Field_...

View on HN

Yes. I did a project ETL'ing data that used the ASCII field and record separators once. It was a joy to deal with compared to CSV.

This is definitely the Right Thing, but how well did other software, such as text editors, handle or render the field and record separator characters? That you can’t directly enter these characters on a keyboard would be a minor annoyance.

The file I used was being generated by a partner company's ERP system so I didn't really have to make any files myself. Having said that, I use a lot of Notepad++ on Windows and using the ALT-029, ALT-028, etc, syntax generated the right codes for direct keyboard entry. They're displayed by Notepad++ as inverse text "RS", "GS", etc.

Is there any reason a 'csv' file can't use E.G.

\\002 2 0x02 STX (start of text)

OR

\036 30 0x1E RS (record separator)

as the record separator instead of a comma? The big reason CSV uses commas is to be human readable and editable, but otherwise...

Ha, we used to do this in the 90s with double pipes || and triple pipes |||.

IIRC PHP BB used || for data separators. The abstraction often leaked.

Reply on news.ycombinator.com