[WIP] fixes in stdlib for -d:checkAbi

[timotheecour]

[DO NOT REVIEW but discussion welcome]
#13926 needs to be merged first

TODO

• requires fixing sizeof for importc enum is wrong: 1 instead of cint.sizeof=4 #13927
• requires fixing CT sizeof(T) is wrong if T or any member is empty object #13945
• make sure CI passes (right now linux 32bit still has issues)
• figure out what's the best way to deal with backward compatibility for the changes involved (maybe simply .since, not sure yet)
• some types are clearly wrong and should simply be fixed
• other types are platform specific edge cases and maybe there's a solution based on an analog of incompleteStruct but for primitive types, eg: .unknownSize:

Ino* {.importc: "ino_t", header: "<sys/types.h>", unknownSize.} = cuint

which would cause CT sizeof+friends to fail for these

• similar to proposal 5 (importc types: CT sizeof, $, {.sizeof.}, const CVAR {.importc.}: int RFCs#205) we could get the "real" sizeof from calling backend compiler on header files; this should work really well, but need to figure out what happens for cross compilation.

relevant comments to address from #13926

• CT sizeof(+friends) for {.importc, completeStruct.} types, enable ABI static checks #13926 (comment)
• CT sizeof(+friends) for {.importc, completeStruct.} types, enable ABI static checks #13926 (review)

distinct not handled by -d:checkAbi

type
  Time* {.importc: "time_t", header: "<time.h>".} = distinct bool

[alaviss]

Please fork this file into posix_linux_i386, or you can adapt posix_linux_amd64 to work for 32bit. I'd say posix_other should just be kept as a template for future OSes.

Also following Linux's definition is not wise, especially since this file is used by many other OS that doesn't have their own dedicated version.

Ideally we need a tool to auto-detect these types (complementary to tools/detect) so that contributors can add support for other POSIX OS faster.

[stale]

This pull request has been automatically marked as stale because it has not had recent activity. If you think it is still a valid PR, please rebase it on the latest devel; otherwise it will be closed. Thank you for your contributions.