Previously, I had been pushing pings to blo.gs, blogrolling, technorati, and weblogs. I removed all of these, and added pubsubhubbub.appspot.com. This involved two steps.
Adding the following to my feed:
<link rel="hub" href="http://pubsubhubbub.appspot.com/"/>
And, adding the following to my publishing flow (based on pubsubhubbub_publish.py):
#!/usr/bin/python import urllib, urllib2 hub = 'http://pubsubhubbub.appspot.com/' feed = 'http://intertwingly.net/blog/index.atom' try: data = urllib.urlencode((('hub.mode','publish'), ('hub.url',feed))) response = urllib2.urlopen(hub, data) except urllib2.HTTPError, e: if not hasattr(e, 'code') or e.code != 204: raise
Scott: I don’t have any hard data, beyond the following entry in my log:
64.233.172.20 - - [04/Mar/2010:14:58:51 -0500] "GET /blog/index.atom HTTP/1.1" 200 21534 "-" "AppEngine-Google; (+http://code.google.com/appengine; appid: pubsubhubbub)"
I also know that sending pings to multiple places slowed down my publishing (technorati alone often took tens of seconds and even would occasionally time out). Hopefully, there is enough critical mass around PuSH now, so that those that crave real-time updates will fetch the data from there.
On a somewhat related note, I just added code to ping the hub when there is a change in my overall comments feed. I have not yet decided what I want to do on my comments feed per post.
I don’t see any “Link” headers in this blog’s http responses... How would a subscriber know the Hub URL or Topic URL? I would think [link] is a Topic URL, but it doesn’t expose a PubSubHubbub Link...
Maybe I don’t get it. That’s how PSHb works, isn’t it? I haven’t seen the Link header in any of the supposed Publishers (Etsy, Getglue, Tumblr etc). The starting point should be the Pub, right? Not the Hub?