• Posts
  • RSS
  • ◂◂RSS
  • Contact

  • External Comment Integration Rewrite

    July 11th, 2012
    tech  [html]
    I've made several improvements to the way comments on other sites about these blog posts show up:
    • They load in parallel: one site being slow won't delay the others.
    • They're cached: I make at most one request per page per 5-minute period so that it doesn't hammer origin sites when I have a popular post.
    • It can pull in reddit and lesswrong posts.
    Technical details:

    When I first added comments it was just for Facebook, and I couldn't run any server side code. So I wrote a simple stateless wsgi app that ran on dotCloud which would fetch comments and return javascript which would document.write them into the post. This would delay rendering of anything below until the request had gone through dotCloud to Facebook, back to dotCloud, and then back to the browser, but there wasn't much of anything below the comments so it didn't much matter.

    A little later I added Google Plus comments, and I didn't rewrite the wsgi app to make requests in parallel, so it didn't even start getting the Google Plus comments until it finished with Facebook, and it didn't display any comments until it heard back from both. Worse, if anything went wrong with either, comments didn't display at all. It also didn't do any caching, so when I had a popular article my little dotCloud wsgi app would get swamped and no one would see comments.

    I've been doing some crossposting to lesswrong lately, and wanted comments to show up the same way as they do with Google Plus and Facebook. I decided to fix the code at the same time. The general idea now is for each site we:

    1. Make a div.
    2. Inject some javascript that:
      1. Makes an XmlHttpRequest to /wsgi/json-comments/<service>/<token&gt
      2. Gets back data in json as:
        [[author, author_link, anchor, comment],
         [author, author_link, anchor, comment], ...]
      3. (Example: /wsgi/json-comments/gp/XhFcxeshNje.)
      4. Writes out the comments into the div.

    The wsgi code handling requests to /wsgi/json-comments/* is mostly the same as before, except that now a single request is only ever for one service, it returns simple json instead of javascript to write out the comments, and it does caching.

    Setting up caching was pretty simple: memcache does almost everything for me. I had to:

      # On Ubuntu at least this also starts memcached running and sets it
      # to run on boot.
      sudo apt-get install memcached python-memcache
    
    And then add a bit of code to my wsgi app:
      import memcache
      import time
    
      # connect to memcached
      mc = memcache.Client(['127.0.0.1:11211'], debug=0)
    
      # example key: "gp/XhFcxeshNje"
      key = "%s/%s" % (service, token)
      t_and_comments = mc.get(key)
      if t_and_comments:
        t, comments = json.loads(t_and_comments)
      else:
        t, comments = 0, []
    
      # cache for 5min
      if time.time()-t > 5*60:
        comments = generate_comments(service, token)
        mc.set(key, json.dumps([time.time(), comments]))
    
      # ideally we could serve stale data while kicking off a background
      # thread, but we're not that sophisticated.
      return comments
    

    Update 2012-07-11: I just modified this to be smarter. It now does two requests, first to /wsgi/json-comments-cached and once that completes to /wsgi/json-comments. The first request is "give me whatever comments you have cached, and don't waste time on any external services!" while the second is "if you have fresh comments cached give me those, otherwise I'll wait for you to go get them."

    Update 2013-03-15: I noticed that comments on notes were no longer coming in through the facebook comment api, so I scraped them and added a mode where my comment code reads from a file. I haven't used notes since 8/2011 and they're not getting comments anymore, so archiving them seems like a fine compromise.

    Update 2013-09-18: The server-side code is on github.

    Comment via: google plus, facebook

    Recent posts on blogs I like:

    More on the Deutschlandtakt

    The Deutschlandtakt plans are out now. They cover investment through 2040, but even beforehand, there’s a plan for something like a national integrated timetable by 2030, with trains connecting the major cities every 30 minutes rather than hourly. But the…

    via Pedestrian Observations July 1, 2020

    How do cars fare in crash tests they're not specifically optimized for?

    Any time you have a benchmark that gets taken seriously, some people will start gaming the benchmark. Some famous examples in computing are the CPU benchmark specfp and video game benchmarks. With specfp, Sun managed to increase its score on 179.art (a su…

    via Posts on Dan Luu June 30, 2020

    Quick note on the name of this blog

    When I was 21 a friend introduced me to a volume of poems by the 14th-century Persian poet Hafiz, translated by Daniel Ladinsky. I loved them, and eventually named this blog for one of my favorite ones. At some point I read more and found that Ladinsky’s …

    via The whole sky June 21, 2020

    more     (via openring)


  • Posts
  • RSS
  • ◂◂RSS
  • Contact