←back to thread

Ask HN: Any active COBOL devs here? What are you working on?

234 points _false | 4 comments | 18 Jul 25 13:05 UTC | HN request time: 0.957s | source

COBOL legacy systems in finance and government are somewhat of a meme. However, I've never actually met a single person who's day job is to maintain one. I'd be curious to learn what systems are you working on?
Show context
mtmail ◴[18 Jul 25 13:48 UTC] No.44604614[source]▶
Met one close to retirement who worked on a ERP system in the food processing industry. Nightly batch jobs would trigger orders from their suppliers, customer service would enter new orders. Two SAP migrations already failed, costing the company millions. All company process knowledge was in code, database fields have been repurposed (but no renamed, too much work), feature development stop long time ago. In parallel a new system was built in-house (no longer trusting external consultants) and his job was explaining what the system does. Probably well paid but he didn't seem to care, he just wanted to work less and retire on good terms.
replies(2): >>44604667 #>>44607954 #
1. abdullin ◴[18 Jul 25 13:53 UTC] No.44604667[source]▶
I grew to like migration projects like that.

Currently working on migration of 30yo ERP without tests in Progress to Kotlin+PostgreSQL.

AI agents don’t care which code to read or convert into tests. They just need an automated feedback loop and some human oversight.

replies(1): >>44605225 #
2. datpuz ◴[18 Jul 25 14:45 UTC] No.44605225[source]▶
I would argue that they need heavy human oversight
replies(1): >>44606344 #
3. Cthulhu_ ◴[18 Jul 25 16:07 UTC] No.44606344[source]▶
For sure; I'll believe that an AI can read and "understand" code, extract meaning and requirements from it, but it won't be the same as a human that knows requirements.

Then again, a human won't know all requirements either; over time, requirements are literally encoded.

replies(1): >>44606921 #
4. abdullin ◴[18 Jul 25 16:48 UTC] No.44606921{3}[source]▶
In systems like that you can record human interactions with the old version, replay against the new one and compare outcomes.

Is there a delta? Debug and add a unit test to capture the bug. Then fix and move to the next delta.