Hacker News Reader

Comment by diego | Hacker News Reader

If you're serious about building a Bitcoin service, you don't need an extra layer between you and bitcoind (or bitcoinj). If you find it too hard to deal with the blockchain on your own via those apis, you probably should think twice about developing a Bitcoin service.

View on HN

I put together balancebadge.com, a simple bitcoind related app. If I didn't use the chain.com API I would have had to run a server with bitcoind (~20GB of data), a server that indexes data from bitcoind (recall that a simple address balance is not exposed in bitcoind), and stores the data in an alternate database (~200GB). There is no way that I would have built balancebadge if I had to setup and maintain 3 servers with 100s of gigabytes of data.

Also, should developers who don't want to run a PostgreSQL server think twice about building web apps? I believe that general purpose software that requires server maintenance and state management should be run as a service. App developers should be focusing on the details of their product –not how to keep general purpose infrastructure alive and well.

edit Full disclosure: I am one of the creators of chain.com

Does one really need to keep the whole blockchain to perform basic things like tracking account balance?

I'm not really savvy on the details, but from the overall understanding of how Bitcoin works I believe one should receive data, keep it until consensus is confirmed (blocks had matured for good, so unless the network went into really weird split-brain situation no "valid" branches would grow out from those blocks), then aggregate and throw data away.

Keeping the whole transaction history up to genesis block seems nearly pointless to me.

Not sure whenever bitcoind can do those, though. Seems that it keeps the whole blockchain, at least for seeding purposes.

And, indeed, one would still need a database and running daemon that'd leech data from Bitcoin network. So, no way to have a badge on static HTML site. That's valid use case.

I wholeheartedly disagree.

Chain.com is something I have been hoping to see for quite some time now, personally.

I run a "Bitcoin service", it calculates the future difficulty level[0], it's just not all that important to require the constant running of a full bitcoind node.

After I completed the Bitcoin version using the blockchain.info API, I realized there was literally no stable API for Namecoin and was forced to place the namecoin difficulty estimator on hold.

The minute that Chain.com implements their namecoin API, I will be subscribing.

Not because I don't have the competance to build and run a crypto coin daemon 24 hours a day, 7 days a week. I simply know that it's a waste of resources for my current needs.

[0] https://cryptanalys.is/difficulty/bitcoin_btc.php

For me, it defeats the vision of Bitcoin as being decentralized. I know some high level API is of warrant nowadays, but we need something that's independent from someone's profits from it.

If you're serious about building a prototype as fast as you can, you don't need to build something someone else has built for you. If you find it to hard to imagine using that service in the long term, you might be right, but you can just replace it if your product goes anywhere.

you don't need to build something someone else has built for you.

Good point, but there are local tools that appear to provide the same functionality as this site: https://en.bitcoin.it/wiki/API_reference_(JSON-RPC) http://bitcore.io/

bitcoind does not index addresses. It only (if you use the `txindex` switch) indexes transactions. You can't query arbitrary addresses using the bitcoind RPC API.

If you find the Bitcoin protocol so difficult that you need to use bitcoind, you should probably think twice about developing a Bitcoin service.

Why? Or is this a "no true Scotsman?"

Because you're paying a huge price for a small amount of convenience: you are trusting a third party so you don't have to deal with a lightweight server and a few gigabytes of data.

It's not like they are providing access to the Twitter firehose or the Google Search Index, which you couldn't have otherwise.

Edit: people are really downvote-happy on HN lately. You should not put a reasonable comment in the negative just because you disagree.

Well for me I can try some ideas without going all-in to setting up a bitcoind server. blockchain.info seems to have a half-hearted attempt at an API too.

> Because you're paying a huge price for a small amount of convenience: you are trusting a third party so you don't have to deal with a lightweight server and a few gigabytes of data.

Can you point me to a ready-to-use address indexer for the blockchain, that can run on a lightweight server and only consumes a few gigabytes of data?

Reply on news.ycombinator.com