https://doc.rust-lang.org/std/time/struct.SystemTime.html

Some of the current model types are based on the "packed" representation in misskey.
There are pack functions that take parameters, and the form of the packed representation returned depends on the parameter.

https://github.com/syuilo/misskey/blob/ec4d5857d80938758ce64930159be2c941d4e30f/src/models/repositories/user.ts#L147-L154

Currently, we do not take these parameters into account, and we deal with it by using Option to fields that depends on the parameter. Since this approach misses quite a lot of invariants, it is expected to introduce a model type that statically considers the pack parameter.

ref: https://github.com/misskey-dev/misskey/blob/5f985ee832eed61e31ceb51eaa1c51810ad6de39/packages/backend/src/services/chart/charts/active-users.ts#L26-L36

before:

{
  local: {
    users: number[],
  },
  remote: {
    users: number[],
  },
}

after:

{
  readWrite: number[],
  read: number[],
  write: number[],
  registeredWithinWeek: number[],
  registeredWithinMonth: number[],
  registeredWithinYear: number[],
  registeredOutsideWeek: number[],
  registeredOutsideMonth: number[],
  registeredOutsideYear: number[]
}

• pages/ #65
• admin/
• games/
• auth/
• room/
• federation/

I'm sorry if this is an obvious question and not even really a bug report but there must be something I'm missing (especially because I don't use misskey).

I'm making a site that generates embeds for Mastodon, Pleroma, and Misskey, for Discord, and Misskey could very much benefit from it seeing as its embeds don't even show the post contents at all. The library I chose doesn't seem to support misskey and it would be a large undertaking for it to do so, and an equally large undertaking would be dealing with ActivityPub which would seemingly be just as hard as just supporting the sites individually, so i'd like to use this in conjunction with it.

What I'm missing is:

If I have https://misskey.bubbletea.dev/notes/8s27fg0gok, 8s27fg0gok being the id, and the usage for get_note is get_note(&self, Id<Post>), and all the implementations for Id can only have numbers fields...how am I supposed to get the note from that id?

If nothing else it would be nice to see an example for get_note in the docs.

Calckey adds a few new endpoints to Misskey, most notably the recommended timeline. It can be called in the exact same way the social/local/global timelines can, and instances have a recommendedTimelineEnabled property.

• We only support fixed set of root-level message type (subnote, channel, api, etc..)
• typingOnMesssage and typingOnChannel are added in v12.71.0
• It is nice to support arbitrary message type to support them and other future extensions

• Save API responses and use it for testing
• This can bring back per-version / ID method misskey-api testing to the stable CI

Currently Aid, Meid and ObjectId exposes their (parsed) fields publicly. If we can stop exposing them we can rely on the format of string representation to limit the range of these fields (currently datetime method can panic if their values are constructed directly c5de018). This is a breaking change.

Currently, misskey-util implements pagination with an endpoint that takes a pagination parameter and returns an array (i.e. Vec). charts/ endpoint differs from the others in that the response is not a single array, but is divided into several fields, so pagination with misskey-util is currently not supported.

You can't have more than two connections to the "shouldShare" channel.

https://github.com/syuilo/misskey/blob/572e475b394aab0fd752783da8648b4f096653fd/src/server/api/stream/index.ts#L220-L222

Quoting the current behavior from the documentation:

Note that currently it is not possible to have multiple connections to the main stream from the same client. If you try to do so, the Future returned by this method will not complete.

This is an inconvenient and confusing behavior for a library. It is expected that the implementation will make the connection the first time and reuse it for the second and subsequent connections.

Unresolved questions

• Where to implement:
  • misskey-websocket: Almost looks good, but feel like the obligation (to duplicate the connection) is misplaced
  • misskey-util: We may need to stop using the extension trait, which is undesirable

ApiErrorId is just a newtype of String.

However, it seems that Misskey's error ID is actually a UUID.
Implementing ApiErrorID with Uuid has a big advantage, especially in terms of the ability to create constants.
It would be very useful to have known error IDs as constants.

Unresolved questions

• Are all Misskey's error IDs UUID?
• Where to implement error ID constants?

e.g. https://docs.rs/misskey/0.2.0/misskey/model/user/struct.User.html

The possible cause is that docsrs flag is not propagated to misskey_api docs while building misskey docs.

Currently, The latest version flag is enabled as a default.

The latest one (12-63-0) is enabled as a default.

This way would force users who specify a version flag to specify default-features = false and enumerate other flags that have nothing to do with the version specification. It could also lead to frequent breaking changes, as the default behavior changes whenever the version of Misskey supported by misskey-rs is increased.

So, it is better to stop including the version flag in the default feature set and tell users to enable the flag for the corresponding version additionally in the documentation.