How to implement PubSubHubbub

How to implement PubSubHubbub

PubSubHubbub is a free, open and decentralized protocol. It relies on webhooks to push feed updates in real-time from publishers to subscribers (feed readers).

Most importantly, PubSubHubbub builds on existing infrastructure: implementing it won’t change or break your current polling infrastructure, and if for some reason something fails, you can still resort to polling (like you probably already do).

Milions of feeds are PubSubHubbub enabled. Let’s use this blog’s feed as an example. It works the same for any other feed.


The first time you fetch the feed, look for <link> element with rel=hub in it 1.

<feed xmlns="">

  <title>Superfeedr Blog : Real-time cloudy thoughts from a super-hero</title>
  <link href="" rel="self" type="application/atom+xml"/>
  <link href="" rel="alternate" type="text/html"/>
  <link rel="hub" href="" />

The most important line is:

<link rel="hub" href="" />

It tells us that this feed is available in real-time at this community hub.

The second most important line is:

<link href="" rel="self" type="application/atom+xml"/>

It tells us the canonical URL for this feed. It’s the actual feed URL that we should use for subscriptions. Most of the time it will be this URL that you might have fetched, but sometimes, it may differ and when that happens, you need to use the self URL.


Here comes the meat if this tutorial on how to implement PubSubHubbub. PubSubHubbub is based on webhooks, which means you need to have an HTTP server able to handle requests coming from the web (not behind a firewall, and no localhost).

In our case, we have a webhook located at this address.

The subscription request is a POST HTTP request. You can issue it using a command line tool like curl or any other type of HTTP library and client. The subscription request is sent to the hub URL (see discovery above).

curl -X POST <hub>
-d'hub.topic=<feed self url>'

When the request hits the hub, the hub will proceed to the verification of intent. This is a step to prevent spam as the hub needs to confirm that you own (and control) the webhook2.

The verification request is a GET HTTP request sent to the webhook which includes all the details of the subscription and an additional hub.challenge parameter. To confirm, the webhook needs to serve a 200 status and output the hub.challenge in the response body.

  • If the subscription is confirmed, the hub will respond with a 204 status.
  • If not, the hub will respond a 422 status and yield an error message in the response body.

Try it yourself:

curl -X POST '' -d'hub.verify=sync' -d'hub.topic=' -d'hub.callback=' -d'hub.mode=subscribe' -D-

It should yield something like:

HTTP/1.1 204 No Content
X-Powered-By: The force, Luke
PubSubHubbub-Version: 0.3
Date: Tue, 10 Feb 2015 13:42:25 GMT
Connection: close


Once the subscription is confirmed, the hub will inform you of any change in the feed by sending POST requests your webhook.

The body of the POST request includes the new content: this is a fat ping. The notification will only include the new entries in the feed, along with the header part of the feed: its title, links… etc.

Many languages and framework will assume that POST requests are the results of forms and will try to parse the content and may show and empty body. If you’re using PHP, use php://input instead of $_POST. With Ruby On Rails, you need to use request.raw_post in your controllers… etc.

Here’s what the body of a notification will look like:

<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="">
    <link rel="hub" href="" />
    <link rel="self" href="" />
    <link title="Publisher example" rel="self" href="" type="application/atom+xml" />
    <title>Publisher example</title>
    <entry xmlns="" xml:lang="en">
        <title>Hello World</title>
        <content type="text">PubSubHubbub is easy!</content>
        <link title="Hello World" rel="alternate" href="" type="text/html" />

Tools and Good Practices

If you’re looking for a feed that updates frequently to test your implementation, we suggest you check this simple application and its feed. It supports PubSubHubbub and updates that you post should be propagated in real-time.

If you’re trying to debug HTTP calls and inspect their content, check out WebhookInbox. It lets you create webhooks that support PubSubHubbub. As an exercise, replace the webhook in the examples above with a RequestBin URL.

Later, you should think about using different callbacks for every feed. It simplifies debugging and handling of notifications by letting you immediately know which feed is linked to which notification. It’s often as easy as adding a query string like ?feed=key at the end of a generic webhook.

You should submit a hub.secret to the hub when issue subscriptions. This secret should be unique and will let the hub sign notifications so that you can easily make sure the notification is coming from the hub, and not from someone trying to impersonnate it. You should also use HTTPS both for the hubs and your callback URL.

Last, but not least, you should know that Superfeedr can act as a default hub for any feed whether they support PubSubHubbub or not. We also normalize the data in JSON or Atom for your convenience…

Happy real-time hacking!

  1. You could also look at the HTTP headers for a Link header (but this is more common for JSON resources), as per the most recent PubSubHubbub spec

  2. Since we used the hub.verify=sync param, the whole process is synchronous which means that the hub will issue the request to the webhook before responding to the subscription request. You could also use hub.verify=async, but it’s much harder to debug, because the hub will not be able to yield an error message if the verification failed. 

Liked this post? Read the archive or

On the same topic, check 2 ways to use superfeedr, http raw body and ping me i'm famous.

Previously, on the Superfeedr blog: Who fetches our blog feed.