…they want their web tech back!
As you probably know, I’ve been busy with a bunch of nasty last-minute bugs that forced me to push out the spring cleaning maintenance patch several times.
I am very happy to say that this release is finally out. It also went fairly smooth and while I have received several bug reports already, all of those are fairly trivial and I will address them over the course of the week.
That said, it wasn’t without its challenges to get to this point. One seemingly trivial bug that was popping up for several people recently was that logo uploads stopped working. The cause of this was in the codebase for a long time, but more people have started to use AI tools to generate logos now and when those AI spit out images, they typically are huge - both in dimensions and file size. The latter is what caused the issue, because while the web framework would read the request and respond with an error based on the size header alone (it didn’t have to parse the request body), the web server would simple abort the request because it still had megabytes of logo data to stream. I expected this to be a config issue, but it turns out that the embedded web server we used so far is just a bit…dogmatic about it. So I switched to another one that allows to “swallow” the remaining request body after a response has been generated.
So far so good…the problem was that this new server dealt with one of the (seriously) legacy web technologies we use in a different way than the old one did. So it took me many hours to find a setup to make it work again. And it re-instilled the motivation in me to get rid of these really old parts of the game (any path that starts with /action/...) as soon as possible. Their tech is close to 30 years old and they’ve become a serious hindrance.