Missing iconv implementation prevents charset conversion in compiled programs

[Faintdawnn]

When using w64devkit, I encountered an error related to missing iconv implementation:

cc1.exe: error: no iconv implementation, cannot convert from UTF-8 to GB

Context:
This error occurs when trying to use charset conversion flags like -finput-charset=UTF-8 -fexec-charset=GBK to control the encoding of string literals in compiled programs.

Root Cause:
w64devkit's GCC build does not include libiconv, which is required for:

• Converting source file encodings (-finput-charset)
• Converting string literal encodings in compiled binaries (-fexec-charset)

Note: This does NOT affect compiler diagnostic messages (errors/warnings), which are determined by terminal locale settings.

Impact:

• Cannot compile code with mixed source encodings
• Cannot control output encoding of string literals for non-UTF-8 environments
• Limits compatibility with legacy systems expecting specific encodings (e.g., GBK on Chinese Windows)

Comparison:
Some older MinGW distributions include iconv support, making these flags functional.

Workarounds:

1. Use UTF-8 for everything (source files and runtime output)
2. Manually handle encoding conversion in code (e.g., using Windows API like MultiByteToWideChar)
3. Switch to MinGW-w64 distributions that include iconv (WinLibs, MSYS2)

Suggestion:
Would it be possible to either:

1. Include iconv support in future builds, or
2. Document this limitation in the README?

This would help users understand the charset conversion limitations when working with non-UTF-8 environments.

Thanks for this excellent minimalist toolkit!

Note: Written with AI assistance, may contain inaccuracies.

[skeeto]

Thanks for your report. This has come up before: #68. I wasn't happy with the way GCC links libiconv, and since this is a rather esoteric feature I decided to skip it. I've added a note to README.md (a5cc1b2).

[Peter0x44]

Well, we might as well throw up a branch that has it for those who want. I'll do it soon.