Here's a little progress report.
Many mechanics have gone from conceptual mess, to testing ideas and finding ways to implement them to clear blocks with description and distribution of functions. Today, it has translated some functions into universal ones. The work is painstaking and tedious. But you do. Apparently, tomorrow the ships learn not only to fly to the target area, but also precisely to it. Whatever the situation and position of the planet. This is by the way a rather difficult task, because the planets do not stand still, but fly in their orbits. And because we have a real simulation -- they're flying at an ever-changing speed, tilting ellipse, etc.
A few days ago, the shooting mechanics started. Projectiles fly to the target, hits are fixed, effects appear. During the test hung under 1200 shells at the same time. It went well. This is considering that the computer runs 2 clients and a server at the same time. If I wanted to, I could start counting hits and all that...potatoes.
One of the invisible but very important changes is the creation of a single class of ship. So far, I've used functional dummies in each individual level, or simply put, game mode. Some for fighters, others for flights on the system, etc. It was a dead end and it was known, but at the time it was faster and easier. This has now been replaced by a single ship, with a clear distribution of data structures, light internal logic, and a whole state structure through which level handlers will know what to do with it, where to move, etc. And they already do. I'm just shoveling a planetary system handler.
He also added an algorithm to search for possible star jumps on a star map. In fact, in the previous version, these transitions were drawn 2 times more. Each suitable pair of stars was connected 2 times. Back and forth, which is not good. Everything's working right now. In its infancy, the algorithm for finding a path to the goal. Also, now you can watch transitions only in a given radius, and not throughout the cluster.
The whole process, at seeming ease, is very laborious, as expected. Unfortunately, the progress is slowed down by the fact that I am still studying the engine. For example, it was possible to make projectiles fly at the right speed in the right direction and record collisions in about a day and a half. Although the idea and the scheme of implementation was thought out in advance, small errors, a huge number of options in the settings, turned the problem for a couple of hours into a problem for almost 2 days. Tests, joint searches, scratching the back of the head. It’s all right! : they eat away most of the time.
The good thing is that despite the slower pace, there is still progress and it is very significant. In the future, the pace will only increase, as I not only get to know the development system better, but the most conceptual and complex things are being done right now. Now the whole architecture is set, the whole skeleton of the game, which then needs to be wrapped in meat and painted. There will be difficult moments, but now almost all the places where I did not know how, but I have to do it, have been overcome. It remains to implement instances of procedural levels, a hexagonal planet, and the most difficult in it has already been done and is working, competent network replication. I don't see anything insurmountable, but I'll have to sweat.
Today I looked at the YouTube review of strategies for 2021. As many as 2 toys that make one person. And it looks good. Large, and medium-sized studios need to smoke to think about why a solo without a budget saws better than their teams. ()). Motivated )