I've introduced a ticket system. There's a link to the ticket page on the comics' info pages now. I didn't yet apply any CSS on it so it looks basic and I know it. I've passed on on the idea of setting up a ticket system before but that was when I still had a lot fewer comics listed than currently, and I've finally all the pieces in place to set up a proper interface for crawler maintenance. Before this, I had pretty much expected to just do the manual update steps with hand written SQL and all even with a ticket and it would've mattered little whether I'd've read about an issue from an email or a ticket.
I only implemented ticket submissions as the first step. There's no way yet to view your submitted tickets. Nor is there any interface for me to view them either, other than by doing a database query. When I add one, removals of dead comics should be much easier to do. Other maintenance operations will still have to wait for later development goals to fall in place.
Sending me email about crawler issues has been a lamentably haphazard business. I'm afraid I'm not going to go through my old emails at this point. If you've sent me one about a comic that still hasn't been fixed, please go and open a ticket. I'm not yet promising more timely fixes than with the old way but at least it'll be much better organised from now on.
I'll be looking into ways to make the ticket system to bring the errors the crawler itself finds to my attention in a more actionable way. It really shouldn't be so that my users would need to ask me crawler issues this much when the crawler could work with me better in the first place.
I spotted an issue with parallel database accesses causing deadlocks with the web site. When that got triggered, the web server would freeze for some time and trying to access the site would just give a gateway timeout. Nasty stuff, I'm hoping that I've fixed it now or at least made the situation a lot less likely. It had to do with the unread comic counts (the numbers next to the check updates link) and how they got altered by using a redirect link. I don't know how many of you encountered that bug but the fix's been in since last Wednesday. The old site used a single database handle for all backend operations (other than AJAX endpoints) and it wouldn't even have run into a situation like this.
I've split the web site code and its authentication layer to separate projects. I'm hoping that the latter will find more users as it's always better for code health to have other people rely on it. I've switched over to using Gitlab for Piperka's code. It pleases me that they're open source, though I'm still somewhat on the fence about git itself. I still have a fondness for darcs even though I used git this time around. If you think you've found a bug or have an development idea then head to the issues page. I've opened a few myself already. Please don't use it to report about crawler bugs with singular comics. If anyone's interested in giving a ride to the code themselves then I'm ready to provide a copy of the database schema.