3 ms·
The most robust way is not to index. Then you can't go wrong, but it is slower. People seem to be ok with it since the ML part takes longer -- for typical codeb
by esafak 1y ago
The most robust way is not to index. Then you can't go wrong, but it is slower. People seem to be ok with it since the ML part takes longer -- for typical codebases, at least.
https://news.ycombinator.com/item?id=44106944 https://news.ycombinator.com/item?id=44106944
- jsnell 1y agoI don't see how indexing is related to this? The question is about how to get the LLM to reliably apply the edit it wants to make. Even when the full current version of the file is in the context, this is one of the flakiest bits of the current LLM workflows.
- esafak 1y agoIn my mind the concern was about the LLMs mental model of what the files look like, which affects edits. I see where you're coming from too.