Tests may not guarantee correctness, but they are often suggestive.
When I have a bunch of regression tests for *real* issues, I feel a lot more confident. I'm covering ways I've failed in the past.
miniblog.
Related Posts
2
It's so hard to talk about probabilities. Casual conversation often conflates 'the likelihood of an event happening' with 'how strongly I believe in my assessment'.
For example, I'm very confident (say 90%) that the coin in my pocket is 50% likely to land on heads.
Generated Code Generates Overconfident Coders: https://www.deeplearning.ai/the-batch/issue-180/
A study of programmers found that using a LLM for completion produced buggier code but users were more confident in it.
I wonder if this generalises to other completion techniques?
Docker Hub has this lovely feature where *they* provide builds of Dockerfiles. For example: https://hub.docker.com/r/etsy/hound/
This is great for making software more trustworthy. I can review the code on GitHub and be confident that I'm running the same code. More platforms should do this.