Compare commits

..

6 Commits

Author SHA1 Message Date
fiatjaf
fd3fa2e75c rename to NIP-FF and broaden the definition of NUDs (everything is a NUD). 2024-11-11 21:48:00 -03:00
fiatjaf
7e34041997 NIP-71: NUDs 2024-05-02 11:33:34 -03:00
fiatjaf_
218fbb1cc7 NIP-54: decentralized wikis (#787)
* draft of NIP-34: decentralized wikis.

* add merge requests.

* add merge request flow

* update nip number

---------

Co-authored-by: Pablo Fernandez <p@f7z.io>
2024-05-02 11:04:55 -03:00
Leo Wandersleb
88246c2741 Require tags to have at least one string
fixes #1193
2024-04-30 22:33:30 -03:00
fiatjaf
bad8826211 nip34: simplify r tag for earliest unique commit. 2024-04-29 14:37:40 -03:00
fiatjaf
243b286582 nip46: signer should fill in pubkey, id and sig on sign_event. 2024-04-25 20:07:38 -03:00
8 changed files with 162 additions and 56 deletions

2
01.md
View File

@@ -56,7 +56,7 @@ To prevent implementation differences from creating a different event ID for the
### Tags
Each tag is an array of strings of arbitrary size, with some conventions around them. Take a look at the example below:
Each tag is an array of one or more strings, with some conventions around them. Take a look at the example below:
```jsonc
{

24
11.md
View File

@@ -60,30 +60,6 @@ The relay MAY choose to publish its software version as a string attribute. The
Extra Fields
------------
### Virtual Relays
Some relays want to expose "virtual" addresses to the same underlying dataset or other types of weird functionality. A basic example is a relay that may serve all notes under its `/` path, but only Italian notes under its `/it` path and only Japanese notes under its `/jp` path. In this case clients may use the virtual (`/jp` etc) paths when browsing the relay, for example; but in other situations, like when selecting relays for building a feed for some specific profile, clients should just want to connect to `/`.
```json
{
"virtual_path": {
"is": true,
"canonical": "/",
"alternatives": [
{"path": "/jp", "description": "only stuff in Japanese"},
{"path": "/it", "description": "only stuff in Italian"},
{"path": "/fr", "description": "whatever"},
{"path": "/pt", "description": "blergh"}
]
},
...
}
```
- `is`: this is `true` when the current path queried is a virtual path.
- `canonical`: this is the path that should be used when the client doesn't care about the virtual.
- `alternatives`: optionally, if a relay wants to announce its virtual paths, this may be useful sometimes.
### Server Limitations
These are limitations imposed by the relay on clients. Your client

5
34.md
View File

@@ -23,8 +23,7 @@ Git repositories are hosted in Git-enabled servers, but their existence can be a
["web", "<url for browsing>", ...], // a webpage url, if the git server being used provides such a thing
["clone", "<url for git-cloning>", ...], // a url to be given to `git clone` so anyone can clone it
["relays", "<relay-url>", ...] // relays that this repository will monitor for patches and issues
["earliest-unique-commit", "<commit-id>"] // usually root commit but a recent commit for forks
["r", "<earliest-unique-commit-id>"] // so clients can subscribe to all events related to a local git repo
["r", "<earliest-unique-commit-id>", "euc"]
["maintainers", "<other-recognized-maintainer>", ...]
]
}
@@ -32,6 +31,8 @@ Git repositories are hosted in Git-enabled servers, but their existence can be a
The tags `web`, `clone`, `relays`, `maintainers` can have multiple values.
The `r` tag annotated with the `"euc"` marker should be the commit ID of the earliest unique commit of this repo, made to identify it among forks and group it with other repositories hosted elsewhere that may represent essentially the same project. In most cases it will be the root commit of a repository. In case of a permanent fork between two projects, then the first commit after the fork should be used.
Except `d`, all tags are optional.
## Patches

7
46.md
View File

@@ -61,8 +61,9 @@ nostrconnect://<local-keypair-pubkey>?relay=<wss://relay-to-connect-on>&metadata
"method": "sign_event",
"params": [json_stringified(<{
content: "Hello, I'm signing remotely",
pubkey: "fa984bd7dbb282f07e16e7ae87b26a2a7b9b90b7246a44771f0cf5ae58018f52",
// ...the rest of the event data
kind: 1,
tags: [],
created_at: 1714078911
}>)]
}),
"tags": [["p", "fa984bd7dbb282f07e16e7ae87b26a2a7b9b90b7246a44771f0cf5ae58018f52"]], // p-tags the remote user pubkey
@@ -121,7 +122,7 @@ Each of the following are methods that the client sends to the remote signer.
| Command | Params | Result |
| ------------------------ | ------------------------------------------------- | ---------------------------------------------------------------------- |
| `connect` | `[<remote_user_pubkey>, <optional_secret>, <optional_requested_permissions>]` | "ack" |
| `sign_event` | `[<json_stringified_event_to_sign>]` | `json_stringified(<signed_event>)` |
| `sign_event` | `[<{kind, content, tags, created_at}>]` | `json_stringified(<signed_event>)` |
| `ping` | `[]` | "pong" |
| `get_relays` | `[]` | `json_stringified({<relay_url>: {read: <boolean>, write: <boolean>}})` |
| `get_public_key` | `[]` | `<hex-pubkey>` |

2
51.md
View File

@@ -32,6 +32,8 @@ For example, _mute list_ can contain the public keys of spammers and bad actors
| Simple groups | 10009 | [NIP-29](29.md) groups the user is in | `"group"` ([NIP-29](29.md) group ids + mandatory relay URL) |
| Interests | 10015 | topics a user may be interested in and pointers | `"t"` (hashtags) and `"a"` (kind:30015 interest set) |
| Emojis | 10030 | user preferred emojis and pointers to emoji sets | `"emoji"` (see [NIP-30](30.md)) and `"a"` (kind:30030 emoji set) |
| Good wiki authors | 10101 | [NIP-54](54.md) user recommended wiki authors | `"p"` (pubkeys) |
| Good wiki relays | 10102 | [NIP-54](54.md) relays deemed to only host useful articles | `"relay"` (relay URLs) |
## Sets

106
54.md Normal file
View File

@@ -0,0 +1,106 @@
NIP-54
======
Wiki
----
`draft` `optional`
This NIP defines `kind:30818` (a _parameterized replaceable event_) for long-form text content similar to [NIP-23](23.md), but with one important difference: articles are meant to be descriptions, or encyclopedia entries, of particular subjects, and it's expected that multiple people will write articles about the exact same subjects, with either small variations or completely independent content.
Articles are identified by lowercase, normalized ascii `d` tags.
### Articles
```js
{
"content": "A wiki is a hypertext publication collaboratively edited and managed by its own audience.",
"tags": [
[ "d", "wiki" ],
[ "title", "Wiki" ],
]
}
```
[INSERT NORMALIZATION EXAMPLES]
The content should be Markdown, following the same rules as of [NIP-23](23.md), although it takes some extra (optional) metadata tags:
- `title`: for when the display title should be different from the `d` tag.
- `summary`: for display in lists.
- `a` and `e`: for referencing the original event a wiki article was forked from.
One extra functionality is added: **wikilinks**. Unlike normal Markdown links `[]()` that link to webpages, wikilinks `[[]]` link to other articles in the wiki. In this case, the wiki is the entirety of Nostr. Clicking on a wikilink should cause the client to ask relays for events with `d` tags equal to the target of that wikilink.
### Merge Requests
Event `kind:818` represents a request to merge from a forked article into the source. It is directed to a pubkey and references the original article and the modified event.
[INSERT EVENT EXAMPLE]
### Redirects
Event `kind:30819` is also defined to stand for "wiki redirects", i.e. if one thinks `Shell structure` should redirect to `Thin-shell structure` they can issue one of these events instead of replicating the content. These events can be used for automatically redirecting between articles on a client, but also for generating crowdsourced "disambiguation" pages ([common in Wikipedia](https://en.wikipedia.org/wiki/Help:Disambiguation)).
[INSERT EVENT EXAMPLE]
How to decide what article to display
-------------------------------------
As there could be many articles for each given name, some kind of prioritization must be done by clients. Criteria for this should vary between users and clients, but some means that can be used are described below:
### Reactions
[NIP-25](25.md) reactions are very simple and can be used to create a simple web-of-trust between wiki article writers and their content. While just counting a raw number of "likes" is unproductive, reacting to any wiki article event with a `+` can be interpreted as a recommendation for that article specifically and a partial recommendation of the author of that article. When 2 or 3-level deep recommendations are followed, suddenly a big part of all the articles may have some form of tagging.
### Relays
[NIP-51](51.md) lists of relays can be created with the kind 10102 and then used by wiki clients in order to determine where to query articles first and to rank these differently in relation to other events fetched from other relays.
### Contact lists
[NIP-02](02.md) contact lists can form the basis of a recommendation system that is then expanded with relay lists and reaction lists through nested queries. These lists form a good starting point only because they are so widespread.
### Wiki-related contact lists
[NIP-51](51.md) lists can also be used to create a list of users that are trusted only in the context of wiki authorship or wiki curationship.
Forks
---------
Wiki-events can tag other wiki-events with a `fork` marker to specify that this event came from a different version. Both `a` and `e` tags SHOULD be used and have the `fork` marker applied, to identify the exact version it was forked from.
Deference
---------
Wiki-events can tag other wiki-events with a `defer` marker to indicate that it considers someone else's entry as a "better" version of itself. If using a `defer` marker both `a` and `e` tags SHOULD be used.
This is a stronger signal of trust than a `+` reaction.
This marker is useful when a user edits someone else's entry; if the original author includes the editor's changes and the editor doesn't want to keep/maintain an indepedent version, the `link` tag could effectively be a considered a "deletion" of the editor's version and putting that pubkey's WoT weight behind the original author's version.
Why Markdown?
-------------
If the idea is to make a wiki then the most obvious text format to use is probably the mediawiki/wikitext format used by Wikipedia since it's widely deployed in all mediawiki installations and used for decades with great success. However, it turns out that format is very bloated and convoluted, has way too many features and probably because of that it doesn't have many alternative implementations out there, and the ones that exist are not complete and don't look very trustworthy. Also it is very much a centralized format that can probably be changed at the whims of the Wikipedia owners.
On the other hand, Markdown has proven to work well for small scale wikis and one of the biggest wikis in the planet (which is not very often thought of as a wiki), [StackOverflow](https://stackoverflow.com) and its child sites, and also one of the biggest "personal wiki" software, [Obsidian](https://obsidian.md/). Markdown can probably deliver 95% of the functionality of wikitext. When augmented with tables, diagram generators and MathJax (which are common extensions that exist in the wild and can be included in this NIP) that rate probably goes to 99%, and its simplicity is a huge benefit that can't be overlooked. Wikitext format can also be transpíled into Markdown using Pandoc. Given all that, I think it's a reasonable suspicion that mediawiki is not inherently better than Markdown, the success of Wikipedia probably cannot be predicated on the syntax language choice.
# Appendix 1: Merge requests
Users can request other users to get their entries merged into someone else's entry by creating a `kind:818` event.
```js
{
"content": "I added information about how to make hot ice-creams",
"kind": 818,
"tags": [
[ "a", "30818:<destination-pubkey>:hot-ice-creams", "<relay-url>" ],
[ "e", "<version-against-which-the-modification-was-made>", "<relay-url>' ],
[ "p", "<destination-pubkey>" ],
[ "e", "<version-to-be-merged>", "<relay-url>", "source" ]
]
}
```
`.content`: an optional explanation detailing why this merge is being requested.
`a` tag: tag of the article which should be modified (i.e. the target of this merge request).
`e` tag: optional version of the article in which this modifications is based
`e` tag with `source` marker: the ID of the event that should be merged. This event id MUST be of a `kind:30818` as defined in this NIP.
The destination-pubkey (the pubkey being requested to merge something into their article can create [[NIP-25]] reactions that tag the `kind:818` event with `+` or `-`

View File

@@ -66,6 +66,7 @@ They exist to document what may be implemented by [Nostr](https://github.com/nos
- [NIP-51: Lists](51.md)
- [NIP-52: Calendar Events](52.md)
- [NIP-53: Live Activities](53.md)
- [NIP-54: Wiki](54.md)
- [NIP-56: Reporting](56.md)
- [NIP-57: Lightning Zaps](57.md)
- [NIP-58: Badges](58.md)
@@ -82,6 +83,7 @@ They exist to document what may be implemented by [Nostr](https://github.com/nos
- [NIP-96: HTTP File Storage Integration](96.md)
- [NIP-98: HTTP Auth](98.md)
- [NIP-99: Classified Listings](99.md)
- [NIP-FF: NUDs](ff.md)
## Event Kinds
| kind | description | NIP |
@@ -171,6 +173,7 @@ They exist to document what may be implemented by [Nostr](https://github.com/nos
| `30402` | Classified Listing | [99](99.md) |
| `30403` | Draft Classified Listing | [99](99.md) |
| `30617` | Repository announcements | [34](34.md) |
| `30818` | Wiki article | [54](54.md) |
| `31922` | Date-Based Calendar Event | [52](52.md) |
| `31923` | Time-Based Calendar Event | [52](52.md) |
| `31924` | Calendar | [52](52.md) |

17
ff.md Normal file
View File

@@ -0,0 +1,17 @@
NIP-FF
======
NUDs
----
`draft` `optional`
This NIP defines the creation of **NUD**s: _Nostr Unofficial Documents_, which are descriptions of standards and sub-standards that do not pertain to this NIPs repo.
Anyone can create a NUD and they are immediatelly valid as soon as they are published. NUDs have owners and can be modified by these owners at any time, but they can also be forked or reinterpreted by others.
NUDs MUST declare the event kinds they use and then those kind reservations SHOULD be taken into account in the NIPs big table of known kinds, with a link to the NUD.
Multiple forks of a NUD can exist at the same time -- although eventually it's natural that one of them becomes the _de facto_ winner. The NIPs repo SHOULD make reasonable judgements about which is which when considering what NUD to add to its big table (it's also ok to have multiple forks added, if there aren't clear winners).
The NIPs repo SHOULD make reasonable judgements about what NUDs to actually add to the big table in the first place, as some may not meet basic decency criteria or may not be used at all, so they SHOULD NOT be added, or they SHOULD be removed later if the assumptions change.