I noticed that stopping the Lemmy process with ctrl+c wasnt working
because the activity channel isnt properly closed. This is now fixed.
Later we should also move the channel from static into LemmyContext,
Im not doing that now to avoid conflicts with #3670.
* Denormalize community_id into post_aggregates for a 1000x speed-up when loading posts
* Remove unused index
* Add creator_id to post_aggregates
* Use post_aggregates as main table for PostQuery
* Make post_aggregates the main table for PostView
* Reformat SQL
* Remove SendActivity and Perform traits, rely on channel
These traits arent necessary anymore now that websocket is removed.
Removing them allows us to use normal actix http handler methods
which are much more flexible, and allow using different middlewares
as well as setting response attributes.
* compiling and create post federating
* clippy
* rename methods, join outgoing activities task
* fix api tests
* no unwrap
* conditional compile
* add back getrandom
* make crates optional
* fmt
* add new function build_post_response_deleted_allowed
* PostDelete uses new function build_post_response_deleted_allowed
* RemovePost uses new build_post_response_deleted_allowed function
* code comments about mod or admin flag having other use
* reformat "cargo +nightly fmt --all"
* Try using drone cache plugin
* Try another path
* Include volume
* Fix formatting
* Include fmt
* Exclude cargo dir from prettier
* Don't override cargo
* Just do check
* Add cache key
* Use different cache plugin
* Add clippy
* Try minio
* Add quotes
* Try adding secrets
* Try again
* Again
* Use correct secret formation
* Add back clippy
* Use secret for the root bucket name
* Try drone cache instead
* Add region
* Add path-style option
* Include cargo clippy
* Include everything again
* Fix formatting
* Don't run clippy twice
* Add `allow` statements for tests to pass
* Adjust endpoint to be a secret
* Fix prettier
* Merge & fix tests
* Try to restart the woodpecker test
* Change the ENV var name
---------
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
* change pool fields to parameters for list
* remove my_person_id and admin fields
* Change recipient id to list param
* Remove TypedBuilder from db_views and db_views_actor
* Split activity table into sent and received parts (fixes#3103)
The received activities are only stored in order to avoid processing
the same incoming activity multiple times. For this purpose it is
completely unnecessary to store the data. So we can split the
table into sent_activity and received_activity parts, where
only sent_activity table needs to store activity data. This should
reduce storage use significantly.
Also reduces activity storage duration to three months, we can reduce
this further if necessary.
Additionally the id columns of activity tables are removed because
they are completely unused and risk overflowing (fixes#3560).
* address review
* move insert_received_activity() methods to verify handlers
* remove unnecessary conflict line
* clippy
* use on conflict, add tests
* Add dev profile to strip symbols and disable debug info (ref #3610)
This significantly speeds up builds:
- with strip symbols and debug 0: 43s
- without: 169s
* add comment, no strip symbols
* Check for dead federated instances (fixes#2221)
* move to apub crate, use timestamp
* make it compile
* clippy
* use moka to cache blocklists, dead instances, restore orig scheduled tasks
* remove leftover last_alive var
* error handling
* wip
* fix alive check for instances without nodeinfo, add coalesce
* clippy
* move federation blocklist cache to #3486
* unused deps
* Add workaround for #3102 until cross-compile work complete
* Minor cleanup
---------
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
dull b