file VS rust_dos

https://www.darwinsys.com/file/

This is the authors website. Apparently yeah its not part of GNU utils, I had no idea, I knew it came with most Linux systems so I looked for the Debian package and found the site linked above.

https://packages.debian.org/bookworm/file

https://github.com/file/file/blob/master/magic/Magdir/gentoo https://www.iana.org/assignments/media-types/application/vnd.gentoo.ebuild (among others)

Look into libmagic.

in installations of Linux, file is a command to retrieve some basic information about a file type. Because contrasting to the organization of Windows, the file extension in Linux can be less important for the file's use (even irrelevant), than the actual content.

In this case, it's hitting this set of rules. I would guess it hits both class and public.

But I don't think this is what you meant. I think your question is something more like: how many file formats are there? The Unix file command recognizes several thousand, and Wikipedia has a list. Of course these are only the most common, and it's hard to delineate what counts as a unique file format. (Is YAML one file format or millions?)

The magic number for all ogg files is the same (OggS) and files magic file looks like it decides the mime type based on the codecs used in the file. However, it looks like application/ogg is used only if the file contains Kate (Karaoke and Text) content.

Also the magic number database for guessing the format of a file:

https://www.darwinsys.com/file/

You do not need segment registers much if you stick to the tiny model. Here is someone compiling Rust to a 16-bit DOS COM executable:

https://github.com/o8vm/rust_dos

Not sure what the approach would be for them to expand that to support segments.

In DJGPP there are macros to allow your protected mode application access physical real-mode addresses (like when you want to write to video RAM). I don't know if IA-16 also does something like that, or if they added far/near keywords to the language like old 16-bit C compilers did (at least the ones I used).

Free Pascal has helper-functions to work with segment+offset pointer pairs, also without having to modify the language itself. I think that would work well enough in C, but I guess the old method of adding non-standard keywords was seen as slightly more convenient.

Demo: https://github.com/o8vm/rust_dos

It is probably possible for almost any old platform with some cross-compilation magic, but not anything that will be officially supported as the compiler-makers focus on modern systems. There is for instance an unofficial 16-bit DOS backend for GCC and at least one or two projects to compile Rust to DOS-executables (that I assume use Clang?) (in addition to 32-bit DJGPP(gcc) for MSDOS that I linked to above). Probably are similar projects to target 68k somewhere?