Minimal rename of go-ipfs in 2022Q2

[lidel]

This is part of ipfs/ipfs#470, read that first.

What

IPFS Stewards are renaming go-ipfs to something else ❄️

👉 Feel free to comment in ipfs/ipfs#471 with name (and logo like one in #8958).

Note: we reserve the right to ignore every instance of Boaty McBoatface.

Why

This is part of a wider, long ecosystem epic (see ipfs/ipfs#470) where we clarify that IPFS is a set of interoperable protocols and conventions, and not a specific implementation, like go-ipfs.

tl;dr:

• go-ipfs will stop squatting "IPFS" name
• we want to encourage people to create their own implementations
• makes it easier to think in terms of "IPFS protocol, libraries and specs" and "IPFS implementations" that are built with the former

When

This will take time, but we want to do the basic rename of project / repo (scope tbd) before July 2022

How

The scope is TBD, we are identifying potential breakage in #8958.

Current proposal is to:

🟢 Rename before July

• README and internal docs
• Git repo and github.com/ipfs/go-ipfs/ imports
  • Caveat: this will impact dist.ipfs.io artifact names. We want to limit the blast radius, so will be building dist under both new and old names, that way no external tooling built in the past 7 years will be broken. PoC in feat: dual publishing go-ipfs and kubo distributions#717
• https://ipfs.io/#install – PR: refactor: rename go-ipfs to Kubo ipfs-website#143
• https://docs.ipfs.io – PR: refactor: rename go-ipfs to kubo ipfs-docs#1185 (waiting for 0.14)
  • ensure https://github.com/ipfs/ipfs-docs/tree/main/tools/http-api-docs automation works with the new repo name and the new name is listed on https://docs.ipfs.io/reference/http/api/

👍 👎 🤔 Need analysis and decision

• https://github.com/ipfs/go-ipfs-http-client – tracked in Point users at canonical kubo/client/rpc library #9124
• https://github.com/ipfs/go-ipfs-api – tracked in Point users at canonical kubo/client/rpc library #9124
• JS client library – Create js-kubo-rpc-client library #9125

🛑 Keep or rename at a later date

Things that use go-ipfs name, that we don't plan breaking at this time:

• keep old name at dist.ipfs.io (https://dist.ipfs.io/go-ipfs/ built in addition to the new name – PR in feat: dual publishing go-ipfs and kubo distributions#717)
• Docker (dual publishing)
  • https://hub.docker.com/r/ipfs/kubo/tags/
  • https://hub.docker.com/r/ipfs/go-ipfs/tags/ (legacy)
• package at NPM (https://github.com/ipfs/npm-go-ipfs) – tracked in Rename NPM package to kubo npm-kubo#51

---

WANT TO PROPOSE THE NEW NAME?

👉 Feel free to comment in ipfs/ipfs#471 with name (and logo like one in #8958).

[hello-smile6]

Maybe golang-ipfs?

[hello-smile6]

@aschmahmann Why not?

[ajnavarro]

greenblock 🤣

[BigLep]

2022-05-10 notes:

1. We need to set and communicate timeline
2. Deduplicate/close between this issue and Renaming ipfs implementations 2021/2022 edition ipfs#470 and the corresponding GitHub Discussion.
3. Determine the set of options for voting.

@lidel is owning on getting decisions wrapped up by the week of 2022-05-16.

[Destroyinator69420]

I think go-ipfs is great name, go-ipfs describes perfectly that the program is an ipfs implementation in golang, I was not aware that the name minimizes other implementations such as rust-ipfs or js-ipfs, I think those names are fine too.

All the best
Destroyinator

[bertrandfalguiere]

I think the idea behind it is to open up the ecosystem to multiple implementations even in the same langage. To make go-ipfs a go implementation of IPFS, not THE go implementation of IPFS. Just like for HTTP, many implementations may coexist, for exemple to specialize on different usecases. It would also remind people that IPFS is a protocol, not a software. I think this a first step for IPFS (the protocol), to reach the next level of popularity.

[lidel]

Quick update:

• We are collecting new name proposals in already existing New name for go-ipfs ipfs#471
  • Posted topic on forums and pinned it for 1 month. It was automatically reposed to Matrix/Discord chatter channel, ensuring we get signal from the community (if anyone cares).
• The renaming will happen AFTER 0.13 ships but BEFORE 0.14
  • We will try to limit the breakage across ecosystem by providing releases under BOTH the old and the new name on https://dist.ipfs.io website – PoC tracked in feat: dual publishing go-ipfs and kubo distributions#717
• Soft deadline: within first two weeks after 0.13 ships, IPFS Stewards will have a call to evaluate names proposed in New name for go-ipfs ipfs#471 and make the decision.
  • we will consider how many 👍 each proposal received, but reserve the right to make arbitrary decision due to SEO, UX and Boaty McBoatface realities

WANT TO PROPOSE THE NEW NAME for go-ipfs?

👉 Comment in ipfs/ipfs#471

[lidel]

IPFS Stewards voted on names proposed here to see which names make the majority happy.
We got top 3-4 selected, go-ipfs team will make final decision later this week.

[RubenKelevra]

@lidel good luck! There are some pretty compelling suggestions :)

[lidel]

Summary of community proposals from discussion thread:

Most popular (captured on 2022-06-14):

1. Kubo +22 / -2
2. Aster +15 / -1
3. Orchid +12 / -1

Long tail:

• Grail +9 / -2
• Simsa +9 / -2
• mox +10

IPFS Stewards voted, and discussed results internally. Identified some concerns around names that clash with other "Web3" / "Dweb" projects, and potential problems when it comes to hearing / pronunciation and non-native speakers (namely, "orchid protocol" exists, and "aster" phonemes are.. problematic)

Stewards narrowed it down to three "safe" choices:

• kubo
• grail
• 🍌

We will now consult with historical maintainers and project leaders, the final decision will be announced later this week.

[lidel]

we will be renaming go-ipfs to:

kubo

safe and popular, comes with a suiting Japanese meaning:

久 long time, old story./ 保 protect, guarantee, keep, preserve, sustain, support.

thank you all for participating in the discussion!

[RubenKelevra]

Cool!

Will ipfs-desktop become kubo-desktop?

Will ipfs-cluster become kubo-cluster?

[bertrandfalguiere]

I guess it will depend on the plans for these projects. Does PL want to nudge people to write alternative implementations on ipfs-cluster or not, will it be tightly coupled with kubo or will the ipfs daemon of ipfs-desftop be swappable, etc. Same for ipfs-cluster.

[lidel]

Closing this one as we've finally got kubo on NPM (ipfs/npm-kubo#51) and deprecated the old name.