CSV

New
adopt
First Added:July 23, 2026

CSV (Comma-Separated Values) is the common text format for exchanging tabular data between spreadsheets and systems. We adopt it for small-to-medium interchange where humans may open the file.

Blurb

This RFC documents the format used for Comma-Separated Values (CSV) files and registers the associated MIME type “text/csv”.

Summary

What it is: Line-oriented records with comma-separated fields; optional header row; quoting rules for commas, quotes, and newlines. Documented in RFC 4180 as MIME type text/csv. Implementations still vary; be liberal in what you accept.

When to use:

SituationNotes
Human-editable dumpsSpreadsheet round-trips, one-off exports
Small tablesConfig lists, seed data, report attachments
Lowest common denominatorPartners who cannot take Parquet

When to skip:

  • Large analytic tables (Parquet)
  • Nested or typed structures (JSON, Arrow)
  • Strict schemas and evolution (table formats / databases)

Trade-offs: Universal and simple; ambiguous dialects, weak typing, and poor compression at scale.

Details

Practical Notes

TopicNotes
SpecRFC 4180 (informational); RFC 7111 updates
MIMEtext/csv with optional header parameter
EnginesDuckDB and most warehouses read CSV directly

References