Curious

This was for early career campus role Interview went as such Interviewer came in Asked some variation on shortest path to destination graph question. I initially said something like DFS but he said I should look for another solution. I proposed BFS and he said it was OK and asked the runtime. I fumbled a bit and said O(n * amount of neighbors each node has) which he gave me some pushes and I got to O(n * m) I asked clarifying questions like what happens if we can't reach the destination, as well as identifying some edge cases like cycles and empty graph Coded it up correctly but forgot the edge case that the start node could be the destination node but then coded that up after he mentioned that. He asked a follow up with something about changing the weights. I proposed some solution like if(neighbor == destination) result = min(result, distance + weight) He said it was a bit hacky then I'm not sure if this was another follow up question or just the original follow up question but imagine it as a normal graph. I proposed dijkstras and gave a bit of an overkill solution cause I wrote out the version that finds the shortest path to ALL nodes. He asked me about my adjacency list and if there are better alternative cpp std data structures we can use. I proposed a vector<vector<int>> when I originally used unordered_map<int, vector<int>>m. He asked about the tradeoffs which I said they're pretty similar except one would have to have a fixed amount of space (vector) versus our dynamically resizable map. I think this is wrong too and the answer has probably something to do with hash collisions. Follow up questions about the job etc I asked something like : Is there anything I could improve just for the future. He said my cpp knowledge is a bit weak as is typical from new grads and I could improve on that but I would only be graded for my performance not my cpp knowledge The end.