New ask Hacker News story: Ask HN: How do you coax useful code from AI?
Ask HN: How do you coax useful code from AI?
2 by memset | 0 comments on Hacker News.
I spend a lot of time trying to get copilot and cursor to help me solve tedious problems. I can never get it to give me code that is anything remotely correct (in the "correctness" sense of the word.) I think my problem is that I am "lazy", but I honestly have no idea how to prompt in a non-lazy way. I feel like if I had a specific algorithm in mind, for example, then I could just code it myself. Here is an example: > i am building a distributed queue. i am using hashicorp memberlist to know when new nodes are added or removed. each node has a shard and replica number in its memberlist metadata. i am using hashicorp raft to set up replication. > Currently, my memberlist only has a single member and raft is set up to bootstrap with itself as the only node. What logic do I need to write in the memberlist delegate to ensure nodes are correctly added to their shard? this should be reliable - the cluster should never be left in an unstable state, or a state where we attempt to process membership changes while a node is still in a candidate state. What should I be doing instead?
2 by memset | 0 comments on Hacker News.
I spend a lot of time trying to get copilot and cursor to help me solve tedious problems. I can never get it to give me code that is anything remotely correct (in the "correctness" sense of the word.) I think my problem is that I am "lazy", but I honestly have no idea how to prompt in a non-lazy way. I feel like if I had a specific algorithm in mind, for example, then I could just code it myself. Here is an example: > i am building a distributed queue. i am using hashicorp memberlist to know when new nodes are added or removed. each node has a shard and replica number in its memberlist metadata. i am using hashicorp raft to set up replication. > Currently, my memberlist only has a single member and raft is set up to bootstrap with itself as the only node. What logic do I need to write in the memberlist delegate to ensure nodes are correctly added to their shard? this should be reliable - the cluster should never be left in an unstable state, or a state where we attempt to process membership changes while a node is still in a candidate state. What should I be doing instead?
Comments
Post a Comment