4 ms·
This is about changing the sorting algoritm. No snark needed. It should be trivial to add a test for the current algorithm (whenever it was introduced) and to r
by Quekid5 3y ago
This is about changing the sorting algoritm. No snark needed. It should be trivial to add a test for the current algorithm (whenever it was introduced) and to replace it afterwards. Heck, even keep running both for bisimulation testing.
- kragen 3y agoprobably colin was more interested in booting faster than in improving testing infrastructure if your interests run the other way, that's great, and hopefully you will succeed in improving freebsd's testing infrastructure, but it's hardly a reason for colin not to speed up booting but i don't think i saw anyone claiming that the change should not be made because it wouldn't be adequately tested?
- Quekid5 3y agoI don't think you fully read the post I was replying to. I was just saying that that poster's argument against changing the sorting algorithm was misguided. Quote: > will be reviewed and tested by 200 people before it reaches production I was only saying that this is a bad argument for changing to any simple-ish sorting algorithm. (Because: add tests if you need assurance) Granted, if you're doing super-complicated sorting algos with fallbacks for different sizes, heuristics for detecting semi-sorted input, etc. you might want something more sophisticated than trivial property-based checking.