Skip to main content

​
Feed concept

The activity feed displays posts from users that the authenticated user follows. Posts are ordered by creation time (most recent first) and paginated using cursor-based pagination.

​
How it works

  1. Query retrieves all users the requesting user follows
  2. Fetch posts created by those followed users
  3. Order posts by creation time and post ID (descending)
  4. Return paginated results with cursor for next page

​
Feed query

Retrieve the activity feed for a user. 
curl -X GET "https://api.example.com/api/feed/johndoe?limit=20" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"
Response: 
{
  "Posts": [
    {
      "PostID": 12345,
      "InitiatorID": 789,
      "Caption": "Beautiful sunset!",
      "CreatedAt": "2026-03-04T18:30:00Z",
      "LikeCount": 42,
      "PostMediasLinks": [
        {
          "PostMediaID": 1,
          "PostID": 12345,
          "MediaType": "Image",
          "MediaURL": "https://cdn.example.com/sunset.jpg"
        }
      ],
      "Comments": [],
      "PostLikes": []
    }
  ],
  "Cursor": "eyJkYXRlVGltZSI6IjIwMjYtMDMtMDRUMTg6MzA6MDBaIiwibGFzdElkIjoxMjM0NX0=",
  "HasMore": true
}

​
Cursor-based pagination

The feed uses cursor (keyset) pagination instead of offset pagination for better performance and consistency.

​
Why cursor pagination?

  • Stable performance at scale: Avoids scanning/skipping rows like OFFSET does
  • Consistency under writes: New posts don’t cause duplicates or missing items during pagination
  • Predictable query cost: Each page query has similar performance regardless of depth

​
Request parameters

  • cursor (string, optional) - Encoded cursor from previous response. Omit for first page.
  • limit (int, optional) - Number of posts per page. Defaults to 20.

​
Response structure

  • Posts (array) - Array of post objects with full details
  • Cursor (string, nullable) - Encoded cursor for next page, or null if no more posts
  • HasMore (boolean) - Whether more posts are available

​
Cursor workflow

​
Fetching the first page

To get the first page, make a request without a cursor parameter: 
curl -X GET "https://api.example.com/api/feed/johndoe?limit=10" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"
The response includes:
  • First 10 posts
  • A Cursor value (if more posts exist)
  • HasMore: true (if more posts exist)

​
Fetching the next page

Use the Cursor value from the previous response: 
curl -X GET "https://api.example.com/api/feed/johndoe?limit=10&cursor=eyJkYXRlVGltZSI6IjIwMjYtMDMtMDRUMTg6MzA6MDBaIiwibGFzdElkIjoxMjM0NX0=" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"
The cursor encodes:
  • CreatedAt - Timestamp of the last post from previous page
  • PostID - ID of the last post from previous page

​
Cursor format

Cursors are Base64 URL-safe encoded JSON: 
{
  "dateTime": "2026-03-04T18:30:00Z",
  "lastId": 12345
}
Encoded as: eyJkYXRlVGltZSI6IjIwMjYtMDMtMDRUMTg6MzA6MDBaIiwibGFzdElkIjoxMjM0NX0=

​
Query logic

The feed query uses this WHERE clause for pagination: 
query = query.Where(x => 
    x.CreatedAt < cursorTime || 
    (x.CreatedAt == cursorTime && x.PostID <= cursorId)
);
Results are ordered: 
.OrderByDescending(p => p.CreatedAt)
.ThenByDescending(p => p.PostID)
The API fetches limit + 1 posts to determine if more pages exist, then removes the extra post before returning.

​
Full pagination example

​
Step 1: Request first page

curl -X GET "https://api.example.com/api/feed/johndoe?limit=2" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"
Response: 
{
  "Posts": [
    {
      "PostID": 300,
      "CreatedAt": "2026-03-04T20:00:00Z",
      "Caption": "Latest post"
    },
    {
      "PostID": 200,
      "CreatedAt": "2026-03-04T19:00:00Z",
      "Caption": "Second post"
    }
  ],
  "Cursor": "eyJkYXRlVGltZSI6IjIwMjYtMDMtMDRUMTk6MDA6MDBaIiwibGFzdElkIjoyMDB9",
  "HasMore": true
}

​
Step 2: Request next page with cursor

curl -X GET "https://api.example.com/api/feed/johndoe?limit=2&cursor=eyJkYXRlVGltZSI6IjIwMjYtMDMtMDRUMTk6MDA6MDBaIiwibGFzdElkIjoyMDB9" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"
Response: 
{
  "Posts": [
    {
      "PostID": 100,
      "CreatedAt": "2026-03-04T18:00:00Z",
      "Caption": "Older post"
    }
  ],
  "Cursor": null,
  "HasMore": false
}
No more posts available - pagination complete.
At scale, consider implementing a fan-out strategy where posts are pre-populated into each user’s FeedContent table when followed users create posts. This trades write complexity for faster read performance.

​
Feed generation strategy

The current implementation uses a pull model where the feed is generated on-demand by querying posts from followed users.

​
Current approach (Pull model)

  • Feed is assembled at query time
  • Queries all followed users’ posts
  • Filters and sorts in the database
  • Simple to implement, no background processing needed

​
Alternative approach (Fan-out model)

For higher scale, consider using the FeedContent table:
  • When a user creates a post, insert a FeedContent record for each follower
  • Feed queries simply read from FeedContent (already filtered and sorted)
  • Trades write amplification for faster reads
  • Better for users with many followers and active feeds
FeedContent structure: 
public class FeedContent
{
    public long UserID { get; set; }        // Follower who sees this
    public long FollowerID { get; set; }    
    public long FollowedUserID { get; set; }
    public long PostID { get; set; }        // Post to show in feed
}
This table enables pre-computed feeds but requires background processing to populate entries when posts are created.

Build docs developers (and LLMs) love

Get started for freeTalk to us