ecsbeats
commented
1 year ago We may just use embeddings databases (such as Chroma if we're able, and possibly use the normal semantic search as a supplement. If we need more fined-tuned features (such as Regex matching), we can use a library such as Google's Codesearch. With these two tools, Bleve is not necessary.
An important part of the Context Engine is the ability to do full-text searches with different object types. For example, the Coding API will need to request context from other files in the filesystem and other places in the current file. We will use Bleve for our search and indexing.