Piperka blog

Inferred schedules

Piperka's main page is kind of useless. I've never had any good idea about what to put there and it's only ever had a short introductory message along with a count of the amount of comics I've personally subscribed to. Yet it's what many well established users use as their landing page, either via search engines or directly, only to immediately click away from there to the updates page. Not anymore. I streamlined it and updates.html page is now gone and what was there is now shown on the main page for logged in users. Going to the old updates page will now redirect. Adjust your bookmarks if you will, even though I see that many of you still use a HTTP URL for accessing Piperka.

I've done a bit of log analysis to determine comics' weekly update schedules, whenever they have a stable enough one to tell. I haven't yet used the inferred schedules as an input for the hourly crawler run, it would be nice to have it ping on archives more often when there's an expectation of new content and let them rest at other times. There's no guarantee yet that the crawler would check on the comic when Piperka says that the update should be out. Once I get that part done I'd expect the accuracy of reported update schedules to increase as well.

As of this writing, out of 5326 active listed comics on Piperka 1313 had any updates during the 12 weeks of log history I'm using as data and 668 had schedules I could infer with my algorithm. I didn't use much time at all to try to optimize it so there likely is still room for making it more accurate. I encountered and reported a bug in GHC along the way.

The inferred update schedules are now a part of comics' info pages. If they have updates during the time period but the algorithm failed then there's a message about inconclusive update schedule. If there's no message about a schedule on it then it had no updates during that time.

A more prominent use for the inferred schedules is that they're displayed on the updates page for the comics a user is subscribed to but is up to date with. I suppose this is the most visible change to the part that people use the most on Piperka in quite a while. I'm not one to do any usability studies with the changes I do to my site. Most often I just come up with an idea and if I like it well enough for my own use then I go through with implementing it. Let me know if you don't like seeing Piperka's guesses about upcoming updates and would like to keep the updates page simpler then I can see about adding the option to not show them.

On a personal note, I'm on my annual vacation. Which means that I have a few weeks to give more attention to Piperka. I'm hoping to finish up with a feature or two.

Tue, 30 Jul 2019 15:18:51 UTC