Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
|
|
Adds a view to avoid having to query and join the users & posts table
inside of the application (which isn't particularly convenient in Diesel).
|
|
This makes it easier to integrate with Diesel, for which this is the
expected column name.
|
|
This guarantees that the anonymous user will always exist and have
ID=1.
|
|
|
|
As a first step of getting the concept of users into the database,
this creates a (somewhat involved) migration that moves all existing
"users" into a new users table and updates the post table to reference
it.
This migration is not yet finalised and still needs to be updated with
something to handle the concept of anonymous users (which I want to
keep around).
|
|
After thinking for a little while I actually think the body of a post
may be more relevant when searching for posts than the thread title.
Right now this is just a hunch and we'll have to find out how it
actually ends up working in real life.
|
|
Adds a materialized view to be used for full-text searches that
indexes the tsvector documents for each post.
|
|
Adds a 'sticky' column to threads and rewrites the thread index to
take sticky markings into account when ordering threads.
Stickies are not yet highlighted in any way in the forum overview.
|
|
This implements the classic thread ordering expected by most forum
users (i.e. the thread with the most recent post is at the top).
|
|
As a data model simplification, the text of a thread's top post should
simply also go into the posts table.
|
|
Adds columns for author names and emails to both threads and posts.
"Users" don't (yet? who knows!) exist as independent entity in
Converse, they are simply "inferred" via OIDC providers.
|
|
|
|
This makes it possible to query posts by thread via Diesel.
|
|
|