Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
catgary
5 months ago
|
parent
|
context
|
favorite
| on:
Some uncomfortable truths about AI coding agents
Because the code is the actual thing, tests can only show that the code fails in certain cases, they don’t actually prove the code is correct.
raw_anon_1111
5 months ago
[–]
If you are writing the correct tests that mirror the requirements, why wouldn’t passing tests mean the code is correct?
catgary
5 months ago
|
parent
[–]
Because it doesn’t? Thats why the field of formal methods exists.
raw_anon_1111
5 months ago
|
root
|
parent
[–]
Then that sounds like you aren’t writing good tests…
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
