What is the difference between ipfs.add and ipfs.dag.put?

Question

Using the js-ipfs lib, I'm struggling to find good information regarding the difference between the following commands:

> await ipfs.add('hello world',{cidVersion:1})
{
  path: 'bafkreifzjut3te2nhyekklss27nh3k72ysco7y32koao5eei66wof36n5e',
  cid: CID(bafkreifzjut3te2nhyekklss27nh3k72ysco7y32koao5eei66wof36n5e),
  size: 11,
  mode: undefined,
  mtime: undefined
}
> await ipfs.dag.put('hello world')
CID(bafyreifg3qptriirganaf6ggmbdhclgzz7gncundvtsyrovyzqigm25jfe)

My expectation: CIDs would be the same.

Would appreciate any pointers.

Answer 1

Below is based on ipfs add and dag put from Kubo RPC, but JS-IPFS was based on it and most likely does something similar (only difference may be that dag put in Kubo expects/requires dag-json by default)

• ipfs add is for adding files and directories and representing them as UnixFS. It will chunk bigger files into smaller blocks and represent them as dag-pb or raw bytes (leaves).
• ipfs dag put allows you to operate on IPLD data structures other than files and directories. In Kubo, this command will assume input to be dag-json and will store it as binary dag-cbor.

You can compare produced CIDs at https://cid.ipfs.tech – in your example:

• ipfs add created raw block (because 'hello world' fits in a single block, no need for dag-pb)
• ipfs dag created dag-cbor (because that is the default --store-codec)