PR 23792: is the only PR that's left for ASMAP (needs cmake changes first) - hope to get this into v30.0

• Kartograph tool, based on rpki-client-nix ()
• use tool from bitcoin Core encode and compress it
• multiple people will run the Kartograph tool

• using an old one should be better than /16
• but using a fresh asmap is better
• quality drops off slowly, probably five years or so. A years one is fine
• if we ship one in Core, it should be fine while the version is supported
• /16 is slighly better than random
• there is a healthcheck in the debug logs

in the asmap-data repo we do a collaborative launch multiple people join in does not make it fully deterministic, but getting close to very good results there is a tool to diff the results, but hard part is to interpret the diff

usually 5 people, often with high matches seperate PR for compressed files, needs to ACK

Are you using the "combining unused ranges": no I don't think so, we can look into this. This would make it smaller. using option -f

Urraca joined and is helping out. Speedups and testcoverage.

• Still open questions?
• Might need more research and data
• is there anything that would help make the ASMAP thing easier and more welcome to review
• concern about decay. How can you show this?
• tool that compares the asmap health scores?

• can be addresses before we turn it on by default
• "this might have a big impact on the network": yes this the goal
• once you have more people running in listening node from their home with PCP, this might help
• for block download you can connect all peers, and do it differently after IBD

• we discussed the a couple of different approaches
• the current on the the favored approach. but needs more review

• New contributor to asmap tooling: Julien Urraca
• asmap Github org
• kartograph repo: creates the blob
• blob stored in repo asmap-data
• half life of the data set: Someone guesstimates 5 years. Presenter says we should always be better off than today. That said a malicious blob is much worse than current situation.
• Timestamp mocking in the tool which makes it almost deterministic (Presenter says it was for the past 6 tries him and the other contributor did)
• The diff tooling is available but the diff data doesn't give you much information at all