Caching timeline, toots and media

Rubicon commented

2024-09-25 07:00:44 +00:00

Owner

This is a discussion about caching toots and media. It’s needed feature because caching significantly reduces the "empty" time as the app started.

Rubicon added the

feature

proposal

label 2024-09-25 07:00:44 +00:00

Rubicon commented

2024-09-25 07:32:59 +00:00

Author

Owner

What’s more, We already have a very simple toot cache for hero animation. The new cache system can ease the burden to set the cache.

Rubicon commented

2024-10-14 16:35:38 +00:00

Author

Owner

I think we may need separated:

timeline cache
toot cache and media cache

The toot and media cache can be done with the Web Cache API, as they can be seen as atomic data. The timeline cache is composed data so we may need indexed db to implement it correctly.

Additionally, the timeline cache may need to able to have hole - if we don’t actually fetch the whole missing part of the timeline.

Compatibility:

IndexedDB 2.0: all targets support
Web Cache API: all targets support

I think we may need separated: - timeline cache - toot cache and media cache The toot and media cache can be done with the Web Cache API, as they can be seen as atomic data. The timeline cache is composed data so we may need indexed db to implement it correctly. Additionally, the timeline cache may need to able to have hole - if we don’t actually fetch the whole missing part of the timeline. Compatibility: - IndexedDB 2.0: all targets support - Web Cache API: all targets support

Rubicon commented

2024-10-18 12:26:11 +00:00

Author

Owner

a RPC server is added into the service worker - we can use it to communicate with the service worker.

Rubicon added this to the (deleted) project 2024-10-18 14:52:18 +00:00

Rubicon removed this from the (deleted) project 2024-10-20 12:41:22 +00:00

Rubicon commented

2024-12-26 14:10:58 +00:00

Author

Owner

CachedFetch uses the Web Cache API, is landed as an experiment: 97bd6da9ac

fetchStatus uses it:

 export const fetchStatus = /* @__PURE__ */ new CachedFetch(
   cacheBucket,
   (session: RemoteServer, id: string) => {
     const headers = new Headers({
       Accept: "application/json",
     });
     if (isAccountKey(session)) {
       headers.set("Authorization", `Bearer ${session.accessToken}`);
     }
     return {
       url: new URL(`./api/v1/statuses/${id}`, session.site).toString(),
       headers,
     };
   },
   async (response) => {
     return toSmallCamelCase(
       await response.json(),
     ) as unknown as mastodon.v1.Status;
   },
 );

We will see if this way works out.

`CachedFetch` uses the Web Cache API, is landed as an experiment: https://code.lightstands.xyz/Rubicon/tutu/commit/97bd6da9ac3e4aec29d73d4bf5cae1dc6d513dbf `fetchStatus` uses it: https://code.lightstands.xyz/Rubicon/tutu/src/commit/5742932c8609dee1de254a5f0bee5f957592408c/src/masto/statuses.ts#L11-L30 We will see if this way works out.

Rubicon added this to the v1.1.0 project 2024-12-26 14:12:49 +00:00

Rubicon commented

2024-12-26 14:14:24 +00:00

Author

Owner

Added to v1.1.0, because some works have to do in this version:

Settings: add an option to delete the cache
fetchStatus: check if the result is failed

Added to v1.1.0, because some works have to do in this version: - [x] Settings: add an option to delete the cache - [x] `fetchStatus`: check if the result is failed

Rubicon added the

infrastructure

label 2024-12-26 14:24:28 +00:00

Rubicon modified the project from v1.1.0 to v2.0.0

2025-01-01 13:24:29 +00:00

Rubicon self-assigned this 2025-01-01 14:21:16 +00:00

Rows
Columns

Caching timeline, toots and media #14