CodesetsConvertStr with dest CS_MIBENUM_UTF_16LE creates bigendian text

[rzookol]

case CS_MIBENUM_UTF_16:
case CS_MIBENUM_UTF_16BE:
case CS_MIBENUM_UTF_16LE: <- ?????
// not entirely correct, but better than nothing
case CS_MIBENUM_ISO_10646_UCS_2:
res = CodesetsBase->utf16Codeset;
break;

this code can't work correctly

[jens-maus]

Feel free to send in a PullRequest which fixes what you think might not be correct.

[rzookol]

Maybe i'm wrong but library lacks of endian conversion

CS_MIBENUM_UTF_16BE
CS_MIBENUM_UTF_16LE

produces the same result as the code doesn't have the byte swapping (so also UTF8 and UTF32 are broken).

So the code needs to take care of host endian and destination endian in all multibyte cases. I would use __builtin_bswapXX() but i really don't want to mess with library internals.

[jens-maus]

Well, that's how open source goes: You create a fork, you mess around with other peoples' code and then you submit your proposal as a Pull Request. And in addition to that you prepare a thorough technical description with example code and example output or show what is actually going wrong instead of going into a theorectical discussion ;-)