ADBC for COBOL – modern database access meets 1959

Name: ADBC for COBOL – modern database access meets 1959
Availability: InStock
Author: ianmcook

by ianmcook·Apr 1, 2026·6 points·0 comments

Visit Project View on HN

AI Analysis

●●SolidNiche GemBig Brain

COBOL records and Arrow columnar arrays share surprising memory layout compatibility.

Strengths

•Zero pipeline duplication—COBOL participates directly in modern analytical workflows
•Thin C wrapper exposes ADBC using only COBOL-native types like PIC S9 and COMP-2
•Runnable GnuCOBOL demos for SQLite, DuckDB, and Snowflake included

Weaknesses

•Extremely narrow audience—COBOL developers are rare and enterprise-locked
•No performance benchmarks compared to traditional ODBC or JDBC connectors

Post Description

I'm one of the core developers of ADBC (Arrow Database Connectivity). This started as a joke — what would it take to call ADBC from COBOL? Turns out, not much. COBOL's fixed-layout record tables and Arrow's columnar arrays make a lot of the same bets about how to represent typed values in memory. The post walks through the architecture and what surprised us. Repo with runnable GnuCOBOL demos for SQLite, DuckDB, and Snowflake: https://github.com/columnar-tech/adbc-cobol