The `deleteNotes` function would not correctly handle cases where cascade
deleted notes were from a different user than the initially deleted note.
Changelog: Fixed
Currently translated at 99.8% (1205 of 1207 strings)
Co-authored-by: kazari <6c577a54-aac9-482a-955e-745c858445e3@simplelogin.com>
Translate-URL: http://translate.akkoma.dev/projects/foundkey/foundkey/ja/
Translation: Foundkey/foundkey
Previously due to bad logic it was impossible for cache entries to ever
be evicted. Ideally this should not be done again but should just use redis
but that can be done another day.
Changelog: Fixed
Currently translated at 99.8% (1204 of 1206 strings)
Co-authored-by: kazari <6c577a54-aac9-482a-955e-745c858445e3@simplelogin.com>
Translate-URL: http://translate.akkoma.dev/projects/foundkey/foundkey/ja/
Translation: Foundkey/foundkey
The backend depends on argon2 since it is used for hashing passwords. Since
it is not used for building the package, the dependency is by the backend
package itself.
Some errors in the queue are not properly handled. For example a blocked instance
will cause the respective queue job to be retried. this should of course not happen
and instead the job should be dropped. This is trying to correct that.
Changelog: Fixed
This adds a check for inbox and sharedInbox URLs to be both
valid and also absolute URLs. If the normal inbox URL is invalid,
the actor will be rejected. If the sharedInbox URL is invalid, it
will be ignored.
Changelog: Fixed
Previously, if a property of the `MkTime` component was changed the displayed value
would not update because the necessary values were computed once on setup and not
made `$computed` so they would be re-computed when necessary.
This is an optimized version made by Volpeon that should run faster if
the instance has a lot of notes.
See <https://is-a.wyvern.rip/notes/9habqldl6j> for a comparison of the
EXPLAIN ANALYZE of the old and new queries.
Co-authored-by: Volpeon <git@volpeon.ink>
Reviewed-on: FoundKeyGang/FoundKey#403
- The `timeout` parameter does not exist in `node-fetch`, so the timeout was not
working properly.
- Refactor the User-Agent header to be set in a central place instead of several
different places.
- Refactor more places to use getResult which handles the timeout and everything
else already instead of the normal `fetch` provided by `node-fetch`.
This seems to be a common occurrence with attempted migrations between different Misskey forks.
This is the best solution that I can come up with; someone who knows more bash could probably make this completely automated.
This fixes federating notes with advanced mfm usage, that looked weird/wrong otherwise.
See also c174f23389
Co-authored-by: syuilo <Syuilotan@yahoo.co.jp>
Changelog: Changed