Added public key compression specs (#137)

## What has changed?

I've added detailed specs for the implementation of public key compression and decompression. The specifications detail the use of the following `multiformat` features:

- `multibase`
- `multicodec`
- `unsigned-varint`

`multiformat` is used to ensure that the implementation has as much flexibility and robustness as feasible.

## Why make the change?

The usage of key de/compression is outside the typical usage of public keys and requires a degree of background knowledge to correctly implement. The purpose of this specification change is to provide this needed background knowledge.

Please also see

- https://github.com/status-im/status-go/issues/1937
- https://github.com/status-im/status-go/pull/1990

---

* Added Public key compression specs

* Added recommendation for encoding type of compressed keys

* Added unrecognised words to wordlist

* Add multibase to the wordlist

* Added a basic example of the multiformat EC key compression concept

* Added parsable to wordlist

* Hex is the only Lingua Franca we need

* Language to make pk de/compression SHOULD implement

* Added terms to glossary explaining key de/compression

* Change terminology from compress to serialise

* Added rationale for public key compression

* Added deserialization to the wordlist

* Concise sentence

* Added url to the wordlist
This commit is contained in:
Samuel Hawksby-Robinson 2020-06-24 11:16:45 +01:00 committed by GitHub
parent 5aa330c124
commit e98a9b76b7
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 226 additions and 6 deletions

View File

@ -6,7 +6,7 @@ title: 2/ACCOUNT
# 2/ACCOUNT
> Version: 0.3
> Version: 0.4
>
> Status: Stable
>
@ -39,6 +39,14 @@ This specification explains what Status account is, and how a node establishes t
- [Identicon](#identicon)
- [3 word pseudonym / Whisper/Waku key fingerprint](#3-word-pseudonym--whisperwaku-key-fingerprint)
- [ENS name](#ens-name)
- [Public Key Serialization](#public-key-serialization)
- [Basic Serialization Example](#basic-serialization-example)
- [Public Key "Compression" Rationale](#public-key-compression-rationale)
- [Key Encoding](#key-encoding)
- [Public Key Types](#public-key-types)
- [De/Serialization Process Flow](#deserialization-process-flow)
- [Serialization Example](#serialization-example)
- [Deserialization Example](#deserialization-example)
- [Security Considerations](#security-considerations)
- [Changelog](#changelog)
- [Version 0.3](#version-03)
@ -106,11 +114,11 @@ The node verifies or derives everything else associated with the contact from th
### Terms Glossary
| term | description |
| ---------------- | ----------- |
| privkey | ECDSA secp256k1 private key |
| pubkey | ECDSA secp256k1 public key |
| Whisper/Waku key | pubkey for chat with HD derivation path m/43'/60'/1581'/0'/0 |
| term | description |
| ------------------------- | ----------- |
| privkey | ECDSA secp256k1 private key |
| pubkey | ECDSA secp256k1 public key |
| Whisper/Waku key | pubkey for chat with HD derivation path m/43'/60'/1581'/0'/0 |
### Contact Discovery
@ -213,12 +221,212 @@ All messages sent are encrypted with the public key of the destination and signe
-->
## Public Key Serialization
Idiomatically known as "public key compression" and "public key decompression".
The node SHOULD provide functionality for the serialization and deserialization of public / chat keys.
For maximum flexibility, when implementing this functionality, the node MUST support public keys encoded in a range of encoding formats, detailed below.
### Basic Serialization Example
In the example of a typical hexadecimal encoded elliptical curve (EC) public key (such as a secp256k1 pk),
```text
0x04261c55675e55ff25edb50b345cfb3a3f35f60712d251cbaaab97bd50054c6ebc3cd4e22200c68daf7493e1f8da6a190a68a671e2d3977809612424c7c3888bc6
```
minor modification for compatibility and flexibility makes the key self-identifiable and easily parsable,
```text
fe70104261c55675e55ff25edb50b345cfb3a3f35f60712d251cbaaab97bd50054c6ebc3cd4e22200c68daf7493e1f8da6a190a68a671e2d3977809612424c7c3888bc6
```
EC serialization and compact encoding produces a much smaller string representation of the original key.
```text
zQ3shPyZJnxZK4Bwyx9QsaksNKDYTPmpwPvGSjMYVHoXHeEgB
```
### Public Key "Compression" Rationale
Serialized and compactly encoded ("compressed") public keys have a number of UI / UX advantages over non-serialized less densely encoded public keys.
Compressed public keys are smaller, and users may perceive them as less intimidating and less unnecessarily large. Compare the "compressed" and "uncompressed" version of the same public key from above example:
- `0xe70104261c55675e55ff25edb50b345cfb3a3f35f60712d251cbaaab97bd50054c6ebc3cd4e22200c68daf7493e1f8da6a190a68a671e2d3977809612424c7c3888bc6`
- `zQ3shPyZJnxZK4Bwyx9QsaksNKDYTPmpwPvGSjMYVHoXHeEgB`
The user can transmit and share the same data, but at one third of the original size. 136 characters uncompressed vs 49 characters compressed, giving a significant character length reduction of 64%.
The user client app MAY use the compressed public keys throughout the user interface. For example in the `status-react` implementation of the user interface the following places could take advantage of a significantly smaller public key:
- `Onboarding` > `Choose a chat name`
- `Profile` > `Header`
- `Profile` > `Share icon` > `QR code popover`
- `Invite friends` url from `Invite friends` button and `+ -button` > `Invite friends`
- Other user `Profile details`
- `Profile details` > `Share icon` > `QR code popover`
In the case of QR codes a compressed public key can reduce the complexity of the derived codes:
| Uncompressed | Compressed |
| --- | --- |
|<img src="https://user-images.githubusercontent.com/5702426/80531063-e98fcc80-8991-11ea-8c02-c354b5828d35.png" width="400" />|<img src="https://user-images.githubusercontent.com/5702426/80501933-f4356c00-8967-11ea-87d8-eae18becece9.png" width="400"/>|
### Key Encoding
When implementing the pk de/serialization functionality, the node MUST use the [multiformats/multibase](https://github.com/multiformats/multibase) encoding protocol to interpret incoming key data and to return key data in a desired encoding.
The node SHOULD support the following `multibase` encoding formats.
```csv
encoding, code, description, status
identity, 0x00, 8-bit binary (encoder and decoder keeps data unmodified), default
base2, 0, binary (01010101), candidate
base8, 7, octal, draft
base10, 9, decimal, draft
base16, f, hexadecimal, default
base16upper, F, hexadecimal, default
base32hex, v, rfc4648 case-insensitive - no padding - highest char, candidate
base32hexupper, V, rfc4648 case-insensitive - no padding - highest char, candidate
base32hexpad, t, rfc4648 case-insensitive - with padding, candidate
base32hexpadupper, T, rfc4648 case-insensitive - with padding, candidate
base32, b, rfc4648 case-insensitive - no padding, default
base32upper, B, rfc4648 case-insensitive - no padding, default
base32pad, c, rfc4648 case-insensitive - with padding, candidate
base32padupper, C, rfc4648 case-insensitive - with padding, candidate
base32z, h, z-base-32 (used by Tahoe-LAFS), draft
base36, k, base36 [0-9a-z] case-insensitive - no padding, draft
base36upper, K, base36 [0-9a-z] case-insensitive - no padding, draft
base58btc, z, base58 bitcoin, default
base58flickr, Z, base58 flicker, candidate
base64, m, rfc4648 no padding, default
base64pad, M, rfc4648 with padding - MIME encoding, candidate
base64url, u, rfc4648 no padding, default
base64urlpad, U, rfc4648 with padding, default
```
**Note** this specification RECOMMENDs that implementations extend the standard `multibase` protocol to parse strings prepended with `0x` as `f` hexadecimal encoded bytes.
Implementing this recommendation will allow the node to correctly interpret traditionally identified hexadecimal strings (e.g. `0x1337c0de`).
*Example:*
`0xe70102261c55675e55ff25edb50b345cfb3a3f35f60712d251cbaaab97bd50054c6ebc`
SHOULD be interpreted as
`fe70102261c55675e55ff25edb50b345cfb3a3f35f60712d251cbaaab97bd50054c6ebc`
This specification RECOMMENDs that the consuming service of the node uses a compact encoding type, such as base64 or base58 to allow for as short representations of the key as possible.
### Public Key Types
When implementing the pk de/serialization functionality, The node MUST support the [multiformats/multicodec](https://github.com/multiformats/multicodec) key type identifiers for the following public key type.
| Name | Tag | Code | Description |
| ------------------ | --- | ------ | ------------------------------------ |
| `secp256k1-pub` | key | `0xe7` | Secp256k1 public key |
For a public key to be identifiable to the node the public key data MUST be prepended with the relevant [multiformats/unsigned-varint](https://github.com/multiformats/unsigned-varint) formatted code.
*Example:*
Below is a representation of an deserialized secp256k1 public key.
```text
04
26 | 1c | 55 | 67 | 5e | 55 | ff | 25
ed | b5 | 0b | 34 | 5c | fb | 3a | 3f
35 | f6 | 07 | 12 | d2 | 51 | cb | aa
ab | 97 | bd | 50 | 05 | 4c | 6e | bc
3c | d4 | e2 | 22 | 00 | c6 | 8d | af
74 | 93 | e1 | f8 | da | 6a | 19 | 0a
68 | a6 | 71 | e2 | d3 | 97 | 78 | 09
61 | 24 | 24 | c7 | c3 | 88 | 8b | c6
```
The `multicodec` code for a secp256k1 public key is `0xe7`.
After parsing the code `0xe7` as a `multiformats/uvarint`, the byte value is `0xe7 0x01`, prepending this to the public key results in the below representation.
```text
e7 | 01 | 04
26 | 1c | 55 | 67 | 5e | 55 | ff | 25
ed | b5 | 0b | 34 | 5c | fb | 3a | 3f
35 | f6 | 07 | 12 | d2 | 51 | cb | aa
ab | 97 | bd | 50 | 05 | 4c | 6e | bc
3c | d4 | e2 | 22 | 00 | c6 | 8d | af
74 | 93 | e1 | f8 | da | 6a | 19 | 0a
68 | a6 | 71 | e2 | d3 | 97 | 78 | 09
61 | 24 | 24 | c7 | c3 | 88 | 8b | c6
```
### De/Serialization Process Flow
When implementing the pk de/serialization functionality, the node MUST be passed a `multicodec` identified public key, of the above supported types, encoded with a valid `multibase` identifier.
This specification RECOMMENDs that the node also accept an encoding type parameter to encode the output data. This provides for the case where the user requires the de/serialization key to be in a different encoding to the encoding of the given key.
#### Serialization Example
A hexadecimal encoded secp256k1 public chat key typically is represented as below:
```text
0x04261c55675e55ff25edb50b345cfb3a3f35f60712d251cbaaab97bd50054c6ebc3cd4e22200c68daf7493e1f8da6a190a68a671e2d3977809612424c7c3888bc6
```
To be properly interpreted by the node for serialization the public key MUST be prepended with the `multicodec` `uvarint` code `0xea 0x01` and encoded with a valid `multibase` encoding, therefore giving the following:
```text
fea0104261c55675e55ff25edb50b345cfb3a3f35f60712d251cbaaab97bd50054c6ebc3cd4e22200c68daf7493e1f8da6a190a68a671e2d3977809612424c7c3888bc6
```
If adhering to the specification recommendation to provide the user with an output encoding parameter, the above string would be passed to the node with the following `multibase` encoding identifier.
In this example the output encoding is defined as `base58 bitcoin`.
```text
z
```
The return value in this case would be
```text
zQ3shPyZJnxZK4Bwyx9QsaksNKDYTPmpwPvGSjMYVHoXHeEgB
```
Which after `multibase` decoding can be represented in bytes as below:
```text
e7 | 01 | 02
26 | 1c | 55 | 67 | 5e | 55 | ff | 25
ed | b5 | 0b | 34 | 5c | fb | 3a | 3f
35 | f6 | 07 | 12 | d2 | 51 | cb | aa
ab | 97 | bd | 50 | 05 | 4c | 6e | bc
```
#### Deserialization Example
For the user, the deserialization process is exactly the same as serialization with the exception that the user MUST provide a serialized public key for deserialization. Else the deserialization algorithm will fail.
For further guidance on the implementation of public key de/serialization consult the [`status-go` implementation and tests](https://github.com/status-im/status-go/blob/c9772325f2dca76b3504191c53313663ca2efbe5/api/utils_test.go).
## Security Considerations
-
## Changelog
### Version 0.4
Released // TODO
- Added details of public key serialization and deserialization
### Version 0.3
Released [May 22, 2020](https://github.com/status-im/specs/commit/664dd1c9df6ad409e4c007fefc8c8945b8d324e8)

View File

@ -11,6 +11,7 @@ blockable
BlockByHash
BlockByNumber
blockchain
BLS
BundleContainer
chainId
Changelog
@ -34,9 +35,13 @@ DApp
dapp
dapps
DDoS
de
De
decrypt
decrypted
decrypting
Deserialization
deserialization
deserialized
deterministically
devp
@ -130,7 +135,10 @@ merkle
MessageType
mixnets
multiaccount
multibase
multicasting
multicodec
multiformats
multihash
MVDS
myPrivateKey
@ -141,6 +149,7 @@ oneof
Oskar
PairInstallation
params
parsable
partitionsNum
partitionTopic
peerID
@ -160,6 +169,7 @@ pre
prekey
prekeys
prepend
prepended
prepending
privkey
protobuf
@ -215,8 +225,10 @@ unix
Upgradability
URI
URIs
url
uuid
UX
varint
Volodymyr
Volosovskyi
Vp