1. 10

There is a search engine now that works on comments and stories. It should handle most simple queries like “bsd jcs” or advanced boolean things. The index updates every 20 minutes.

User messaging is implemented. There is a user tree that I made mostly for my own amusement. I might disable this later.

The site has a mobile view that looks decent in Android and iOS. It’s all through CSS screen size queries.

The routine for finding duplicate stories works better now and will find similar ones with http:// vs. https://, trailing slashes, etc. Like Hacker News, there is a 30-day window where it will consider the same URL as just an upvote for the previously posted submission. After 30 days it will allow it to be posted as a new story. Also, Google Analytics URL parameters are stripped out of URLs now. They’re stupid and will hardly ever correspond to how the URL is actually being clicked on (i.e., someone posts a link copied from an e-mail broadcast and the UTM parameter says it’s from an e-mail campaign, yet all of the clicks from lobste.rs are coming through the web).

The other day I tweaked the story sorting algorithm to give a bigger window while we’re still growing here, so that popular stories can stay on the front page for a couple days rather than new stuff always pushing it down the page. As more things get posted, I will shrink this window down a bit.

Story submissions can be previewed now before submitting, which is useful for text posts like this one.

Some other minor bugs have been fixed and styles tweaked. If you find anything weird, just send me a message.

  1.  

  2. 2

    Thank you so much for stripping out the Google Analytics stuff.

    1. 2

      Great work jcs. Site looks great and runs super fast on my Galaxy Nexus. Never really understood why HN never took advantage of CSS screen size queries to shrink down the main content area. Its great not having to rely on a 3rd party app or mobile site to view the content.

      1. 1

        Please consider making a partial to display duplicate stories with date, eg:

        • 2012-08-16 Bit-twiddling to improve bloom filter performance – 8 points, 0 comments
        • 2012-07-08 Bit-twiddling to improve bloom filter performance – 16 points, 22 comments
        • 2012-05-28 8 secrets your bloom filter needs that the liberal media won’t tell you – -3 points, 0 comments
        • 2011-12-17 Bit-twiddling to improve bloom filter performance – 2 points, 1 comment

        It’d be nice both on the Sumbit page to help users decide if they should, and on story pages to connect discussions.

        1. 1

          I like the user tree, now only if I trusted anyone else with the power of the front page.

          1. 1

            Really nice with the search – it will turn out useful as the number of users and stories grow. And thanks for the pink pdf tag marks :)

            1. 1

              jcs, would you care to elborate on the sorting a bit? Will clicking Home give us the highest ranked stories within the last couple of days with same-rank-stories being sorted on date? And how is everything out of the two day range displayed? It doesn’t seem to be sorted by submit time.

              1. 1

                It uses the same algorithm as Reddit for ordering stories. I’ve just changed the window on which it operates to 36 hours. As the site grows it will shrink down to 24 or 12 hours.

                1. 1

                  Is any of this based on the reddit codebase?

                  1. 3

                    No, all custom Ruby on Rails code. I had a desire to rewrite it in Go but as the features have been piled on, it has been getting more complex and looking less likely like that will happen.

                    I used Reddit’s “hot” sorting algorithm for stories, and Evan Miller’s confidence algorithm for comments (which Reddit also uses).