Non-programming tools like 3D model editors are another case though as outside of very expensive suites like Mirai, things have progressed much better nowadays. But again this doesn't apply everywhere: for example Deluxe Paint was an easy and popular tool to use for 2D games in the 80s and early to mid 90s and a modern pixel art editor doesn't provide that much more functionality outside of what you can get with modern resources (e.g. unlimited undo).
You're underestimating the role played by good abstractions in developer productivity.
>For most games, programming is far from the biggest labor cost.
And as you say, the other bigger costs are out of the picture when re-implementing old games. So, programming is the largest remaining part, almost by definition of 're-implement'.
When making a game from scratch, you need to produce art assets (often 50-70% of the cost of the game is art assets alone), design gameplay mechanics, market the game, test the game, and distribute the game (which, in the old days, often meant making physical cartridges or disks). Let's say that 10% of the labor in making a game from scratch went to programming. When you're re-implementing a game, all you need recreate is that 10%. So if re-implementing a game is an order of magnitude easier than making a game from scratch, there's no reason to think that programming games has gotten any easier.
Even if you take out the possibility of grabbing something like SGLIB, the two most common data structures in a game you'll need are lists and dynamic arrays and you can create a generic header for both in an hour.
What sort of abstractions does C++17 enable that you think are useful for creating a 2D sidescrolling platformer (or anything other decades old game) that are so empowering compared to C to sidestep C++17's deficiencies?
Anything is better than C at providing abstractions for building games, because any language offers more than procedures, crude arrays and structs. C++, Rust, C#, Python, Java, Swift, etc. All of these would run circles around C for almost any kind of possible project.
First of all, type safety is not necessary to create something fast which is what the original comment was all about and everything i'm writing is in the context of mmoez writing how modern tools enable individuals to do work that decades ago would need entire teams and my reply to that was that this applies to some types of tools and someone who knows C++17 enough to make such a game like those made decades ago wont gain implementation speed from the language itself as opposed using something like C89. I repeat this context because i have a feeling from your replies that you are taking my comments out of it.
Second, you can use the preprocessor to create type safe lists, dynamic arrays, etc. In fact SGLIB does exactly that, this is why you pass a type to the macros that declares your list types and functions. But outside of SGLIB's approach you can also use include files as templates with parameters passed outside of it. Here is an example i wrote some weeks ago:
https://news.ycombinator.com/item?id=19879059
This creates types and functions based on the parameters you give to the header.
> Anything is better than C at providing abstractions for building games, because any language offers more than procedures, crude arrays and structs.
Also a somewhat decent macro system, but so does C++. However i never argued that C is good at providing abstractions (as i show above you can, but it can be a PITA). C++ and some other languages are better than that, sure.
What i argued about is that those abstractions will not cause someone who has enough knowledge to write a game like the one submitted here in C++17 to make it "that would need teams decades ago" faster than if he'd used something like C89. In other words, considering the type of project we're discussing about, even if he gains any speed, it'd be minimal.
And secondly, C is probably the least productive programming language still in use, because of the huge overhead of working with memory. Don't even bother claiming this is not so, I spend enough time reviewing C code to know that this language should be banned for any new project.
> Don't even bother claiming this is not so, I spend enough time reviewing C code to know that this language should be banned for any new project.
Yeah whatever.
While programming languages can nudge architects in certain directions, this is only true for rather special languages, such as Erlang. Both Java and C++ support standard programming paradigms, can and are used to create many types of architectures.
Recommending C for anything except Linux kernel development in 2019 is a public disservice. If more of us were appropriately harsh towards C, instead of finding excuses for it based on nostalgia the SW world would really be a better place.