It's interesting that you bring up "more code metadata" and "self-hosted deployment" as strengths of Sourcegraph. I work for a company that licenses the self-hosted enterprise SG. We have had bugs open for months due to code metadata issues crippling the usefulness of the tool. Specifically, deleted repositories are not getting de-indexed and archived repositories are not getting indexed as such. This makes search results extremely noisy and in some cases outright incorrect. This forces our teams to create extensive "search contexts" in order to find relevant results. I use this tool daily and advocate for more usage internally, which speaks to its usefulness, but it's painful to teach new users all the caveats of poor data quality.