New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We鈥檒l occasionally send you account related emails.

Already on GitHub? Sign in to your account

鈿★笍 v0.34.0 RELEASE 馃殌 #1721

Closed
alanshaw opened this Issue Nov 19, 2018 · 7 comments

Comments

Projects
None yet
4 participants
@alanshaw
Copy link
Member

alanshaw commented Nov 19, 2018

Ahh yeah, it got fast

馃椇 What's left for release?

馃敠 Highlights

馃弾 Reading Protobuf DAG nodes is super fast

Reading Protobuf DAG nodes just got a serious speed boost as well as a memory reduction. Previously Protobuf DAG nodes (dag-pb nodes for short) carried a multihash property (a CIDv0) with them. This used up loads of CPU time calculating the hash for a given node as it was retrieved from storage and with the advent of CIDv1 there was no guarantee the hashing algorithm and length were correct.

So, we removed it! 馃

Ok so that had the unfortunate consequence of making the object API a whole lot less useful when writing data - the DAG nodes you got back were basically the data you put in. Hence the object API was refactored to return CIDs instead of DAG nodes for write operations...and we all lived happily ever after.

馃殼 Adding content is blazingly fast

We found a big bottleneck (or should I say small bottleneck?) that meant the bigger the file you added to IPFS, the longer it would take - oh no 馃槶! Good news folks, we've seen the speed of adding large files increase by 50% or more in some cases! 馃挜

Read all about it in the pull request notes.

馃巹 HAMT support added to MFS

We're putting the whole of npm on IPFS! These days npm is like 5.3 terabytes of data and it's all going in MFS. This is super rad, but we needed a good sharding strategy because 5TB is a lorra lorra files.

If you're curious about npm on IPFS, check out the project here.

馃摚 IPNS over pubsub and DHT

Get informed of IPNS record updates and read and publish your IPNS records to the DHT. It's all there and it's all awesome. For those of you new to IPNS, let me give you the lowdown - IPNS puts the mutable in immutable 馃ぃ It's an age old problem, content addressing is rad and all that, but if I change something the hash changes - 馃憥 boo...but wait, IPNS solves this, you get a permenant address for changeable content - hooray \o/.

IPNS over pubsub gets the word out quicker to peers that are interested when an IPNS record changes. IPNS over DHT allows peers to find and resolve your IPNS address to some content in the first place! In the next JS IPFS release (0.35) the DHT will be enabled by default and it's going to be epic.

鈿撅笍 CID handling improvements

Smoosh your CIDs into whatever version you like - you can now add data under a version 0 CID and get it back using a version 1 CID and vice versa. Now that you have this freedom you can encode them with whatever multibase encoding you like. Version 0 CIDs are all base58btc but if you convert to a version 1 CID you can encode it with base2, base32, base64url or whatever:

# base2
010111000000010010001000000100011011010100010010000001010010111001110001011010111100010100000111000011101010101010101101111100000001011101110001011110100001000100111010101101010111111001000111110001001010000101100010110000001000011110101110100100010101110110100010110100110000001110
# base32
bafybeicg2rebjoofv4kbyovkw7af3rpiitvnl6i7ckcywaq6xjcxnc2mby
# base58btc
QmT78zSuBmuS4z925WZfrqQ1qHaJ56DQaTfyMUF7F8ff5o
# base64url
uAXASIEbUSBS5xa8UHDqqt8BdxehE6tX5HxKFiwIeukV2i0wO

This is all work to smooth out the eventual transition to base32 CIDv1 by default - a move to allow CIDs to be used as valid URL origins. Have a read of this for more.

To help you craft your artisanal CIDs we've introduced a --cid-base option to a bunch of CLI commands (and ?cid-base option to a bunch of HTTP API endpoints). Check it out:

jsipfs add file.txt --cid-base=base32
added bafybeibns4lrebrxaymvyshgmki5biwh6cd53idxfpen6ysomzrywtsm44 file.txt

馃挭 Deps got upgrades!

We got you covered with WebUI 2.3 (quic support, responsive navbar, and a language selector), libp2p 0.24 and many other upgrades giving you a faster and smaller JS IPFS.

馃彄 DHT bumped to next release

The Distributed Hash Table(tennis) is finally here! This is the biggest leap forwards for JS IPFS since sliced bread. It allows your IPFS node to find peers and content way more easily and it has full interop with Go IPFS nodes so y'all have the full IPFS network at your fingertips 馃挜 I know, you're welcome - send flowers to @vasco-santos.

馃彈 API Changes

  • Object API methods that write DAG nodes now return a CID instead of a DAG node. Affected methods:
    • ipfs.object.new
    • ipfs.object.patch.addLink
    • ipfs.object.patch.appendData
    • ipfs.object.patch.rmLink
    • ipfs.object.patch.setData
    • ipfs.object.put
    • More info
  • DAGNode instances, which are part of the IPLD dag-pb format have been refactored. These instances no longer have multihash, cid or serialized properties. This effects the following API methods that return these types of objects:
  • Files API methods add*, cat*, get* have moved from files to the root namespace. Specifically, the following changes have been made:
    • ipfs.files.add => ipfs.add
    • ipfs.files.addPullStream => ipfs.addPullStream
    • ipfs.files.addReadableStream => ipfs.addReadableStream
    • ipfs.files.cat => ipfs.cat
    • ipfs.files.catPullStream => ipfs.catPullStream
    • ipfs.files.catReadableStream => ipfs.catReadableStream
    • ipfs.files.get => ipfs.get
    • ipfs.files.getPullStream => ipfs.getPullStream
    • ipfs.files.getReadableStream => ipfs.getReadableStream
  • New core files API methods added:
  • DHT API methods renamed and return types changed
    • ipfs.dht.findprovs renamed to ipfs.dht.findProvs and returns an array of PeerInfo
    • ipfs.dht.findpeer renamed to ipfs.dht.findPeer and returns a PeerInfo
    • ipfs.dht.query now returns an array of PeerInfo
    • More info
  • A new option is available in the CLI, HTTP API and core that will allow the multibase encoding to be specified for any CIDs that are returned as strings.
    • A --cid-base option has been added to the following CLI commands:
      • jsipfs bitswap stat
      • jsipfs bitswap unwant
      • jsipfs bitswap wantlist
      • jsipfs block put
      • jsipfs block stat
      • jsipfs add
      • jsipfs ls
      • jsipfs object get
      • jsipfs object links
      • jsipfs object new
      • jsipfs object patch add-link
      • jsipfs object patch append-data
      • jsipfs object patch rm-link
      • jsipfs object patch set-data
      • jsipfs object put
      • jsipfs object stat
      • jsipfs pin add
      • jsipfs pin ls
      • jsipfs pin rm
      • jsipfs resolve
      • Note: these two MFS commands already implement the --cid-base option:
      • jsipfs files ls
      • jsipfs files stat
    • A ?cid-base= query option has been added to the following HTTP endpoints:
      • /api/v0/bitswap/wantlist
      • /api/v0/bitswap/stat
      • /api/v0/bitswap/unwant
      • /api/v0/block/put
      • /api/v0/block/stat
      • /api/v0/add
      • /api/v0/ls
      • /api/v0/object/new
      • /api/v0/object/get
      • /api/v0/object/put
      • /api/v0/object/stat
      • /api/v0/object/links
      • /api/v0/object/patch/append-data
      • /api/v0/object/patch/set-data
      • /api/v0/object/patch/add-link
      • /api/v0/object/patch/rm-link
      • /api/v0/pin/ls
      • /api/v0/pin/add
      • /api/v0/pin/rm
      • /api/v0/resolve
    • A cidBase option has been added to the following core functions:
      • resolve
    • NOTE Using the CID base option in bitswap, dag and object APIs WILL NOT auto upgrade your CID to v1 if it is a v0 CID and WILL NOT apply the encoding you specified. This is because these APIs return IPLD objects with links and changing the version of the links would result in a different hash for the node if you were to re-add it. Also, the CID you used to retrieve the node wouldn't actually refer to the node you got back any longer. Read this for further context.
  • All API methods that returned big.js instances now return bignumber.js instances.

Release Checklist

鉂わ笍 Huge thank you to everyone that made this release possible

By alphabetical order, here are all the humans that contributed to the release:

馃檶馃徑 Want to contribute?

Would you like to contribute to the IPFS project and don't know how? Well, there are a few places you can get started:

鈦夛笍 Do you have questions?

The best place to ask your questions about IPFS, how it works and what you can do with it is at discuss.ipfs.io. We are also available at the #ipfs channel on Freenode.

@daviddias

This comment has been minimized.

Copy link
Member

daviddias commented Dec 9, 2018

This release is going to be HUUUUGEEEE!!

@daviddias

This comment has been minimized.

Copy link
Member

daviddias commented Jan 13, 2019

@alanshaw with this release, can js-ipfs move to base32 by default?

@mkg20001

This comment has been minimized.

Copy link
Contributor

mkg20001 commented Jan 14, 2019

@alanshaw Possibly want to make this release even bigger than it is? I'm currently working on stardust, a replacement for ws-star that actually works. I could ship it behind a flag with some help on this pr #1812 which requires this one to be finished libp2p/js-libp2p-crypto#125

@daviddias

This comment has been minimized.

Copy link
Member

daviddias commented Jan 14, 2019

@mkg20001 additional transports can be added through adding a custom libp2p bundle -- https://github.com/ipfs/js-ipfs#optionslibp2p. No need for special flags :)

@mkg20001

This comment has been minimized.

Copy link
Contributor

mkg20001 commented Jan 14, 2019

@daviddias I'm talking about replacing ws-star with stardust upstream in js-ipfs. So that EXPERIMENTAL.stardust would switch ws-star for stardust. See libp2p/js-libp2p-websocket-star#70

@alanshaw

This comment has been minimized.

Copy link
Member

alanshaw commented Jan 14, 2019

@alanshaw with this release, can js-ipfs move to base32 by default?

I'd love it to but there's no PRs implementing it yet. Possibly in 0.35 or more likely in 0.36.

0.34 will bring js-ipfs up to speed with go-ipfs in terms of CID version agnostic get and be ahead of it with the --cid-base option.

Essentially the rough plan now is:

  • 0.35 - DHT
  • 0.36 - base32 CID v1

@alanshaw Possibly want to make this release even bigger than it is?

Absolutely not! This release needs to leave the building asap. 馃ぃ

@aphelionz

This comment has been minimized.

Copy link

aphelionz commented Jan 15, 2019

Great work everybody! Exciting stuff.

@alanshaw alanshaw closed this Jan 17, 2019

@wafflebot wafflebot bot removed the in progress label Jan 17, 2019

@alanshaw alanshaw unpinned this issue Jan 17, 2019

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment