* Model post and profile aggs in bsky
* Add background, onCommit, db and background stats
* Add aggregations for post and profile
* Use background queue in bsky labeler, misc plumbing
* Process background throughout bsky tests
* bsky tidy and fixes
* Build views using agg tables in bsky
* Views of feedgens on bsky
* hook up indexing
* Setup custom feeds on bsky
* graph service + getList routes
* migration
* Feed views refactor for bsky, implement getFeed w/ auth passthrough
* mute lists
* add getListMutes
* mute list tests
* add mutes table & mute/unmuteActor routes
* Passthrough auth from bsky appview to feedgen, fixes/tidy/tests
* add getMutes
* implement mute functionality
* remove unspecced
* mute tests
* mute impl tests
* remove proxy tests
* Test custom algos on bsky
* add mute state to snapshots
* table & indexing plugin
* getBlocks
* bail out of mute subquery when no viewer
* fleshing out services
* Test pds getFeed proxy
* tidy, fix build
* tidy
* Fix minor race in bsky mutelist tests
* blocks in feed views
* db migration, impl block logic & tests
* Update pds proxy snaps based on appview mutes
* missed some files on last commit
* ensure process after unmute
* update snap
* move adminAuth to seedclient param
* apply blocks in getFeed
* update snaps
* upadate one last snap
* pr feedback
---------
Co-authored-by: Devin Ivy <devinivy@gmail.com>
* Misc notification and optimization in bsky, tidy
* Handle notifs on message queue in bsky appview
* Sync-up profile labeling behavior in bsky
* No async work during actor unindexing in bsky, include additional indexes
* Prevent deeply nested notifs w/ hellthread exception in bsky appview
* Add indexes to bsky appview for more efficient actor deletion/unindexing
* Update pds proxy snaps for updated bsky profile labeling
* Return custom suggestions in bsky appview
* Test out-of-order indexing of aggregations on bsky appview
* Proxy pds getPopular to hot-classic feed on appview. Proxy getPopularFeedGenerators to appview.
* Make did field required on feedgen view
* Misc notification and optimization in bsky, tidy
* Handle notifs on message queue in bsky appview
* Sync-up profile labeling behavior in bsky
* No async work during actor unindexing in bsky, include additional indexes
* Prevent deeply nested notifs w/ hellthread exception in bsky appview
* Add indexes to bsky appview for more efficient actor deletion/unindexing
* Update pds proxy snaps for updated bsky profile labeling
* Return custom suggestions in bsky appview
* Model post and profile aggs in bsky
* Add background, onCommit, db and background stats
* Add aggregations for post and profile
* Use background queue in bsky labeler, misc plumbing
* Process background throughout bsky tests
* bsky tidy and fixes
* Build views using agg tables in bsky
* Views of feedgens on bsky
* Setup custom feeds on bsky
* Feed views refactor for bsky, implement getFeed w/ auth passthrough
* Passthrough auth from bsky appview to feedgen, fixes/tidy/tests
* Test custom algos on bsky
* Test pds getFeed proxy
* tidy, fix build
* tidy
* Model post and profile aggs in bsky
* Add background, onCommit, db and background stats
* Add aggregations for post and profile
* Use background queue in bsky labeler, misc plumbing
* Process background throughout bsky tests
* bsky tidy and fixes
* Build views using agg tables in bsky
* Views of feedgens on bsky
* tidy, fix build
* tidy/fix bsky migration
* Model post and profile aggs in bsky
* Add background, onCommit, db and background stats
* Add aggregations for post and profile
* Use background queue in bsky labeler, misc plumbing
* Process background throughout bsky tests
* bsky tidy and fixes
* Build views using agg tables in bsky
* ✨ Hydrate subject author's repo in moderation reports
* 🚧 Expose only handle of the subject repo on moderation reports
* 🚧 Regenerate lexicons off main
* 🧹 Cleanup unnecessary joins
* ✅ Updating snapshots
* ✨ Make handle optional to allow serving reports even if handle doesn't exist
* ✨ Port over subjectHandle implementation to bsky package
* wip
* first pass done
* chnage around how channels function
* bugfixing
* fix up migration & tests
* fix dev-env
* add prep migration for stream fixing
* application code changes to use new repo_event table
* dont send evts
* add outgoing_repo_seq table + sqlite migrations
* patch up down migration
* patching up a couple things
* update test
* patchup
* re-enable subscription route
* add outgoingSeq column
* updating
* accidental add
* temporarily disable stream
* resolve merge conflict
* add migration
* migration for seq
* fix test
* id -> seq
* get schema lock ids
* tidy
* fix server test
* fix up
* use bsky uniqueLockId approach
* tidy
* gh actions build
* build branch
* build branch
* tweak test
* rm constraint rename
* small bugfix
* check for seq not null
* fix up issues
* one more small change
* tidy
* no build branches
* ✨ Allow searching repo by DID
* ✨ Port over did search from pds
* ✅ Add tests for bsky implementation of admin searchRepos
* 🧹 Cleanup according to PR review
* temp disable hot-classic
* Switch labels in whats-hot to values list
* Switch labels in getPopular (legacy hot-classic) to values list
* Reenable hot-classic, switch labels to values list
* remove branch build in workflow
* nsid: reduce 'name' length from 128 chars to 63 chars
* nsid: remove 'nsid-ns' (glob pattern) from main NSID syntax validation
This special variant syntax should be handled separately. Eg, an
"nsid-ns" should not be accepted as an NSID in Lexicons looking for a
regular NSID.
* nsid: update syntax check to match domain rules
This is to be closer to actual underlying specification needs. For
example, domains are allowed to start with a digit (like 4chan.org), and
can't start *or* end with a hyphen.
Restricts the "name" part further, to be alphabetic only.
Note that starts-with-digit domains could break some lang
auto-generation variable name rules. The docs/spec strongly recommend
against such domains for use with NSID, but it seems incorrect to
restrict at the standard/lexicon level without a clear pan-language
consistent set of rules.
* did-resolver: disallow did:web with path segments
* identifier: reduce max length of DID from 8 KB to 2 KB
* identifier: add DID test with escaped ':' in identifier
* did-resolver: have prettier run on tests/
* identifier: add additional punycode test cases
* identifiers: small tweaks from review
- type in did:web comment
- include actual max DID string length in error message
* xrpc-server tests: remove digits from NSIDs
* identity: fix import and function errors
* xrpc-server tests: remove digits from NSIDs
* xrpc-server: include tests/ in prettier:fix
* xrpc-server: lint fix
* Rename getLabelsForSubjects to getLabelsForUris
* Generic method for getting profiles organized by did or aturi subject
* Batch label fetching for posts and authors in feeds and threads
* Tidy
* To avoid performing many view refreshes in a row w/ different leaders, schedule them upfront
* Apply feed threshold to with-friends and mutuals algos
* Revert increase to getRepo commit chunk size
* Alter feed threshold approach to be tighter and relative to cursor
* ✨ Allow ignoring moderation reports by subject
* 🔧 Remove unwanted package.json change
* ✨ Remove subject format specifier
* ✨ Port over ignore subject implementation to bsky package
* Revert "✨ Port over ignore subject implementation to bsky package"
This reverts commit 3c782c1805548d1ebdbbd08b68b6e92ac97fdee5.
* 🚧 Port over ignore subject implementation to bsky package
* ✨ Generate lexicons for bsky
* Temporarily skip online/valid checks for feedgens while external feedgens catch-up
* tweak algos
* hook up did resolver with cache
* Re-enable whats-hot w/ materialization
* pulling a fast one on the query optimizer
* clean up
* disable branch build
---------
Co-authored-by: Devin Ivy <devinivy@gmail.com>
devin ivy
Daniel Holmgren
Foysal Ahamed
bnewbold