You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Due to too optimistic merging into the master branch by me it is in a broken state and populated with code of high complexity. To me it is very clear what went wrong and how we will avoid it in the future.
The stable branch is a compromise of a more modern client the main problem being incomplete and unsolvable maps.
The last release we had has too few maps and a very short play time so even if we port all the old maps to stable we still don't have much of a game. Meaning we need new maps#82.
Since our main game mechanic is very limited at the moment especially on stable it might be wise to not base all off our content on the teleportation puzzles but to also make exploration part of the experience.
We will roll back the ECS on master in favour of a simpler System that utilises multiple inheritance. #85 Edit: Dorian: Multithread isn't going away, I'll just limit the System/Simulation runner thread count to 1 for now.
It's a soft limit.
Geert: So what was the problem with ECS?
Dorian:
General complexity it has when you must combine multiple components every single time.
Breaking down behaviours is nice but implies you make code to handle that behaviour split.
C++ multiple and virtual inheritance already does that for you at a lesser overhead cost than ECS since it can use platform-specific optimizations through the compiler's implementation.
Due to too optimistic merging into the master branch by me it is in a broken state and populated with code of high complexity. To me it is very clear what went wrong and how we will avoid it in the future.
The stable branch is a compromise of a more modern client the main problem being incomplete and unsolvable maps.
The last release we had has too few maps and a very short play time so even if we port all the old maps to stable we still don't have much of a game. Meaning we need new maps #82.
Since our main game mechanic is very limited at the moment especially on stable it might be wise to not base all off our content on the teleportation puzzles but to also make exploration part of the experience.
We will roll back the ECS on master in favour of a simpler System that utilises multiple inheritance. #85
Edit:
Dorian: Multithread isn't going away, I'll just limit the System/Simulation runner thread count to 1 for now.
It's a soft limit.
Geert: So what was the problem with ECS?
Dorian:
General complexity it has when you must combine multiple components every single time.
Breaking down behaviours is nice but implies you make code to handle that behaviour split.
C++ multiple and virtual inheritance already does that for you at a lesser overhead cost than ECS since it can use platform-specific optimizations through the compiler's implementation.
#55 #66
The text was updated successfully, but these errors were encountered: