Reindex should not try to do everything all at once
Each datagraph type should have its own reindex job (I think)
Reindexing should be based on a persisted "last index" date, matched to the datagraph date
Generally, reindexing shouldn't happen that often, it's something for when things go wrong (weaviate was off for a period of time, semdex was enabled some time after the instance was set up, etc)
Indexing in general needs some work:
titles of posts aren't being indexed for some reason
deletions currently aren't handled
visibility is not handled correctly: #283 only published resources should be indexed, un-publishing should count as deleting from the semdex
Southclaws
commented
1 week ago
Off the back of #285:
Indexing in general needs some work: