I don't know how I got here and ended up watching most of this video, but what the heck is the point of this talk? The question from 26:26 is the whole reason I came here, and the response is basically, "Yea I don't actually know what impact is, but I just know theoretically that it would be more efficient."
I don't think that is a fair summary of his answer to that question. This talk is more about building efficient web based games, as opposed to improving existing ones. Thanks for watching!
@@elementengineering I understand what you are saying that the question was technically about "reimplementing" a game for an A B comparison, but the point I'm trying to make is there is no data or even anecdotes that indicates how "efficient" this strategy of design is. The one question I was hoping to get answered was "What is the tradeoff or impact from designing your applications this way?" I feel like this is a core part of the audience member's question.
@@Gibster It's a good question! I've asked Leigh himself, who replied: The main drawback really is the added complexity if the game design does not feature realtime interactions between multiple entities. For example, a puzzle game is turn based and can be modelled well by a state machine (which can be messy in ECS), and adding new puzzle block types is straightforward with a standard OO paradigm. Implementing this in ECS might require creating many highly specific components and systems for every block type which could be laborious without inheritance and polymorphism and does not really align with the intended use case of modelling more broad concerns like physics, collision detection, etc. Hope that helps!
I don't know how I got here and ended up watching most of this video, but what the heck is the point of this talk? The question from 26:26 is the whole reason I came here, and the response is basically, "Yea I don't actually know what impact is, but I just know theoretically that it would be more efficient."
I don't think that is a fair summary of his answer to that question. This talk is more about building efficient web based games, as opposed to improving existing ones.
Thanks for watching!
@@elementengineering I understand what you are saying that the question was technically about "reimplementing" a game for an A B comparison, but the point I'm trying to make is there is no data or even anecdotes that indicates how "efficient" this strategy of design is. The one question I was hoping to get answered was "What is the tradeoff or impact from designing your applications this way?" I feel like this is a core part of the audience member's question.
@@Gibster It's a good question! I've asked Leigh himself, who replied:
The main drawback really is the added complexity if the game design does not feature realtime interactions between multiple entities.
For example, a puzzle game is turn based and can be modelled well by a state machine (which can be messy in ECS), and adding new puzzle block types is straightforward with a standard OO paradigm. Implementing this in ECS might require creating many highly specific components and systems for every block type which could be laborious without inheritance and polymorphism and does not really align with the intended use case of modelling more broad concerns like physics, collision detection, etc.
Hope that helps!