[jjice]

> he found that ninety percent failed to provide a correct solution after several hours of working on it

I wonder how many university algorithms engineers would run into a similar situation with. I'm not very far out of university, so I have an unfair advantage, but take some engineers with at least 10 year experience and have them implement Dijkstra's by memory. Not saying this is necessarily a bad thing. If you're in industry for years, you have gained a lot more applicable knowledge to your work, kind of like how runners have a different kind of strength and muscle than weight lifers - they both meet their needs.

That said, I'm surprised at 90% overflowing code. Is it all in that midpoint calculation?

[mark-r]

Yes, it's always the midpoint calculation. Everybody forgets that (low+high) could overflow. It's trivial to work around once you know - just replace mid=(low+high)/2 with mid=low+(high-low)/2.