Used addresses

[szist]

Describe your idea, in simple words.

Ability to filter out full addresses based on either delegation or payment credential.

Why is it a good idea?

It could help identifying which addresses are tied to a script (e.g. DEX pool addresses where each pool has a different delegation) or getting a list of addresses from where the delegator ADA is considered.
Secondly, it would help with interacting with HD wallets to detect which addresses were used already.
The sqlite structure and indexes would already help with it and some distinct select on top of the addresses should solve this.

Are you willing to work on it yourself?

Maybe, but I'm not as fluent in Haskell.

[KtorZ]

Ability to filter out full addresses based on either delegation or payment credential.

Hey, I've got good news, you can already do that 😄!
The GET /match/{pattern} endpoint accepts any kind of patterns, including patterns to match on either delegation or payment credentials. Have a look at the user manual on Patterns for more details about the syntax and how to construct patterns.

Secondly, it would help with interacting with HD wallets to detect which addresses were used already.

There's already a field on each match result indicating whether a UTxO has been spent or not as well as a query flag that allows you to filter UTxO based on that criteria. Surely, querying any pattern and requesting only the ?spent one would give you only results that have already been used.

[szist]

All true, post processing is fine as long as the n is a relatively low number. For example once you have active trading accounts and n grows, it gets increasingly costly to post-process the txos.

If you also consider paging, then the list might become incomplete, unless the client fetches everything. And yeah I meant just listing the addresses used with an sql distinct select. It seemed like a low hanging fruit :) an alternative to https://docs.blockfrost.io/#tag/Cardano-Accounts/paths/~1accounts~1{stake_address}~1mirs/get on delegation key side, and no alternative for getting all addresses associated with a scripthash currently :)

[KtorZ]

To be frank, I don't think post processing a list with a single traversal would be a major issue. I am constantly benchmarking against the jpeg.store or SundaeSwap contract addresses which have 100k+ UTxOs. Fetching, streaming and post processing the whole result (150MB+ of data) with jq takes less than 3s. And that's about the biggest request one can do on mainnet at the moment.

Regarding pagination, you're 100% right. I'll drop some thoughts on the other discussion tomorrow.

[szist]

Well, if you want use locally 150m is not that much. Having microservices or using it to send data over mobile network, that's quite a lot of data. :)

[KtorZ]

But Kupo is primarily meant to be used as a local service 😁...

Maybe you could tell more about your use-case?

[szist]

I have a few in mind :D. But I would just leave it for later or to a dm. For one of the first specific case I am looking to replace that exact endpoint I posted there from blockfrost. I don't like relying on such external centralized services. Carp seems too heavy (even though it tries to be modular) , db-sync-based stuff (even just with ledger state not much better).

(second more of a painpoint I can share, blockfrost and explorers lack the ability to get the addresses for a script, which already caused me a few issues elsewhere, as well for some aggregators trying to measure TVL and assuming that all orders/lps/farms would be on the same address not considering that all those utxos could be delegated. There you could have a lighter kupo running to get all addresses for a script and the use blockfrost/other address based service to get the rest )