Doing this would allow a user to request repos as needed. Using a bloom filter as a repo header.
What exactly is the problem that you’re trying to solve?
He just learned about Bloom filters, and really wants to use them.
What if we combined taxis with bloom filters? Doing this would allow users to request taxis as needed. Using a bloom filter as a phone number.
Wha
A bloom filter of what exactly?
What do you mean by “repository” here? Code repositories, like git? ATProto repositories? The Repository code pattern? Something else?
What are you on about?
Bloomfilters intentionally can have false positives but no false negatives. They’re great for searching over very large datasets as a quick and cheap way to drop lots of data from more expensive search operations.
We use a deterministic checksum for this so you know what you’re getting without ambiguity.
Pretty much anywhere you can think of a compelling reason to use a bloom filter, they are nice to use. In this case, though, you aren’t really proposing anything, so it is difficult to say. What repositories? What are you looking for?
If it is some sort of dependency check or similar which only needs a membership check, this would be a good way to filter out a great many. As long as your good with a few false positives that you actually charge, then maybe.
I feel like this is part of a full conversation that we aren’t seeing.
You just… invented dependency management.
You would still need to know what you’re looking for, whether that’s an ID (repo name / full IRL) or some (unspecified part of a) repo header. It’s surprising why someone of your skill level asks such an odd, unspecified question. Why do you being up a random idea and ask the community any that’s good or bad? Shouldn’t you offer some arguments as to why you think it’s good from the start, unless it obvious?
The only advantage I can think of would be to more easily find forks or even the same content on different hosts without having to know / verify possibly differing the repo names. But that brings is to the largest downside: security. Git hosters over namespaced repos with configurable access control. Your idea would be susceptible to impersonation, unless you being signing and some novel way to manage trust and verify signatures. Quite some hassle with little benefits 🤔
Start with a proof of concept distributed file sharing network
