Viktor

It would be very useful if the search engine could be, in part or as a whole, loaded with common crawl data. This would help set up additional instances as...

The search engine is currently limited to English, and actively filters out content in other languages. It would be desirable to support other languages as well. It's a bit unclear...

URLs in API search results are inconsistently encoded. This is likely due to inadequate or incorrect normalization at some earlier stage. URLs are supposed to be unencoded in the API.

In practice, BBR is strongly advised.

Very low quality results. Gibbons' "Decline and fall of the roman empire" is among them, but relatively far down.

The LIVE_CRAWL actor enters a fail state every time the corresponding executor node restarts. Nothing in the logs to explain why. Doesn't appear to happen _while_ it's running, but when...