Thanks! It's using Elastic's Search UI (https://github.com/elastic/search-ui) and Ant Design (https://github.com/ant-design/ant-design).

I think that error was just because the server was overloaded - sorry about that.

It indexes the default branch of each repo.

Cool. Keep up :) definitely gonna share with my co-workers.

Can't wait for filename filters which would make this the perfect solution

Thanks :) If you type into the path filter box, that'll match against the full path for each file, so you can use that to filter on a filename.

There isn't any deduplication, although that will hopefully be less of an issue at this point since there's a limited number of repositories in the index.

It was based on the number of stars/forks and the size of the repository.

There must be something else or something wrong, because you indexed one of my small repo (~100 stars, ~20 forks, ~20Mb) and not the bigger ones (~500 stars, ~100/150 forks, ~150Mb)

Maybe he is limiting it to repositories of 50 MB or less, for example.

Looking around at repositories I'm familiar with this seems to be the case.

or possibly there's a "time decay" element where more recently "popular" repos are prioritized, not just based on absolute start/fork count

Thanks! It's built on top of Solr. It fetches the repos from GitHub - it should pick up any updates to repos within a few days. It's running on a couple servers with 20 cores each, which is not really enough for the traffic it's getting right now.

Have you seen livegrep?

Blazing fast multi-repo regex code search. May be more expensive to run in prod, not sure.

This is so good I imagine you're gonna need more.

Cool！I love it.