Reading through all of that, they're saying that they avoid a jungle of ifdefs to support multiple similar architectures, and they say the way to this is "isomo

Reading through all of that, they're saying that they avoid a jungle of ifdefs to support multiple similar architectures, and they say the way to this is "isomorphism", which is supposed to be the opposite of abstraction. Ok I get the gist, but no examples; how do they achieve that? A different page gives a list of supported calls, which seems to make it clear that what they did was a (probably a good job of) factorization of the problem domain into a set of standardized library calls [1] that model the abilities of the entire set of architectures closely. Applications/games can then use that library without #ifdef'ing for different architectures and still get highly efficient results. Very clever, but of course they don't claim that their approach generalizes to very different architectures, like say X11 on RISC-V or something. [1] https://retroprogramming.iwashere.eu/midres_library:functions https://retroprogramming.iwashere.eu/midres_library:function...