Thread

How does pyramid compare to strfry performance wise? The pyramid readme mentions memory mapping. Is it build in a similar way?

I didn't benchmark, but it should get pretty close to strfry for query operations, it's very inspired by it and also uses LMDB. It also does a noveau thing with direct memory mapping of a huge file with encoded events that I came up with and may be completely broken but may also end up saving some milliseconds. The most important part is that events are deduplicated across the main relay and all the subrelays.

> The most important part is that events are deduplicated across the main relay and all the subrelays. Isn't duplication part of the redundancy? Or is this just aggregation

This looks interesting, especially the idea of kicking people out to prevent bad actors from being added permanently to the system