feat: only-hash option for add/addFromFS/addFromURL

[JonKrone]

This adds only-hash as a potential querystring arg for files.add, util.addFromFS, and util.addFromURL: ipfs.add(<content>, { only-hash: true }).

only-hash allows users to generate a hash for some content without actually adding the content to ipfs.

related issue: #509

needed for js-ipfs: ipfs/js-ipfs#1233

[ghost]

Does this also support -n?

  -n,          --only-hash           bool   - Only chunk and hash - do not write to disk.

[JonKrone]

Does this also support -n?

I was thinking that alias support like that is only available through a CLI, (js-ipfs here). Should this API also support aliases? Please explain if I'm missing something, but would you expect ipfs.add(<content>, { n: true }) to be valid?

[ghost]

Ah I'm sorry, I didn't notice we're in js-ipfs-api

[JonKrone]

Haha, no worries 😉. Thanks for the look-over, though.

[JonKrone]

CI question: Does anyone know why ESLint finds errors but doesn't report them here? This has tricked me a few times, I think.

[ghost]

Should this API also support aliases?

Actually: maybe. go-ipfs supports option aliases on the API, e.g. /api/v0/refs?r=true&arg=QmFoo (because that's where most of the CLI options eventually end up). @diasdavid's call.

[ghost]

But again this is just options objects passed to an API function yeah? Then probably not.

[JonKrone]

Actually: maybe. go-ipfs supports option aliases on the API

Cool, I think js-ipfs could change that as a mild chore across the route handlers later on. Do you know why go decided to? Aliases on a CLI save time and lots of frustration but most code through an API is generated so aliases aren't nearly as helpful.

[JonKrone]

CI is failing on object.patch tests: https://travis-ci.org/ipfs/js-ipfs-api/jobs/345383716#L3541

These tests do not fail locally but are also failing on VMX's most recent PR: #701.

[daviddias]

Alias are supported for some methods (early days) but we decided to bet on the full options names for docs and tests. (Aliases mean + docs and + dup tests)

[daviddias]

@vmx could https://github.com/ipld/js-ipld-dag-pb/releases/tag/v0.13.1 be breaking the tests? HAven't looked too much into it.

[vmx]

@diasdavid: Nope, I've just publish js-ipld-dag-pb 0.13.1 just 30mins ago.

[JonKrone]

Current status: ready for review / merge.

CI failure is pubsub multiple nodes connected: empty() list when no topics are subscribed.

The object.patch CI tests that were failing for vmx and me are passing now for both PRs.

[daviddias]

@JonKrone any reason why to use a different github account for your commits?

See how things are commited by @jkrone rather than @JonKrone

[JonKrone]

@diasdavid No reason. It is not a different Github account, just a local git config user.name setting that was set to jkrone. I've synced the names and pushed a commit that reflects it.

[daviddias]

let's get #705 by @vmx in first to that we can see CI green in every PR :)

[daviddias]

@JonKrone please rebase master onto this branch when you have a chance.

[daviddias]

@JonKrone merging this branch into a branch of the repo itself so that we can collaborate