diff options
| author | 2025-05-15 09:40:48 +0000 | |
|---|---|---|
| committer | 2025-05-15 09:40:48 +0000 | |
| commit | 3cff4b2d7d138f45787a94e6e64acdccb00e8951 (patch) | |
| tree | d18f18289b3082b129d6e129aa383ddde5869521 /internal/headerfilter/filter.go | |
| parent | [chore] Change default database in example config to sqlite, update docs. (#4... (diff) | |
| download | gotosocial-3cff4b2d7d138f45787a94e6e64acdccb00e8951.tar.xz | |
[chore] various federatingdb tweaks (#4178)
after seeing a potential reported federating worker lockup i decided to start digging into the federatingdb code. this PR encompasses:
- removes one of our last unused interface types `federatingdb.DB{}`, replacing it with a struct type `*federatingdb.DB{}`
- in `transport.dereferenceLocal()` differentiates between an unsupported lookup type and ErrNoEntries to reduce unnecessary calls, and reduce potential lockups that may occur while trying to call our own endpoints that then call `federatingdb.Lock()`
- removes a bunch of the locks on follow state changes since the DB already synchronizes that
- removes the unnecessary `pub.Clock{}` struct field and type passed to the transport controller
frankly it would be great if we could remove the locking in `federatingdb.Lock()` and instead handle it ourselves as it gets very confusing trying to figure out what functions will have locks held. but i guess that's one for when we move further away from the go-fed/activity/pub package usage.
Reviewed-on: https://codeberg.org/superseriousbusiness/gotosocial/pulls/4178
Co-authored-by: kim <grufwub@gmail.com>
Co-committed-by: kim <grufwub@gmail.com>
Diffstat (limited to 'internal/headerfilter/filter.go')
0 files changed, 0 insertions, 0 deletions
