We've been using NSIS 2.50 for some time now and I'm trying to update to the newest 3.0x version. I'm using Unicode true. I've ran into an issue, which I'm failing to understand. We use a switch for mapping native language names to language IDs, more or less like this:
${Switch} "${LANGNAME}"
${Case} "${LANGFILE_ALBANIAN_NAME}"
StrCpy $0 "${LANG_ALBANIAN}"
${Break}
${Case} "${LANGFILE_ARABIC_NAME}"
StrCpy $0 "${LANG_ARABIC}"
${Break}
; Other cases
${EndSwitch}
The error I'm getting from compilation:
Bad text encoding: C:\Users\me\AppData\Local\Temp\nst9352.tmp:9
!include: error in script: "C:\Users\me\AppData\Local\Temp\nst9352.tmp" on line 9
Error in macro _EndSwitch on macroline 9
The temporary file is apparently created by LogicLib, which then tries to include it. The file really doesn't have any valid Unicode encoding (I'm posting just a snippet from the file):
!insertmacro _== `$2` `Shqip` _LogicLib_Label_433 ""
!insertmacro _== `$2` `???????` _LogicLib_Label_434 ""
!insertmacro _== `$2` `Catal�` _LogicLib_Label_441 ""
The strings with invalid UTF-8 characters seem to be encoded in various ANSI encodings (some seem to be Western European, some Central European etc.), while the question marks are saved as real question marks (0x3F). ${LANGFILE_NLFID_NAME} is defined in language files as native name using the LANGFILE macro from LangFile.nsh. I looked at the language files and they are encoded in UTF-8 BOM and look all right. So it looks like the native name is re-encoded to ANSI or something for ${LANGFILE_NLFID_NAME}?
I'm pretty sure I'm making some stupid mistake, but I can't really figure out what it is.
Looks like a bug, it will be fixed in the next release.
Since you are not using the fall-through Switch feature you can just use Select or If/ElseIf instead:
${Select} "${LANGNAME}"
${Case} "${LANGFILE_ALBANIAN_NAME}"
DetailPrint LANG_ALBANIAN
${Case} "${LANGFILE_ARABIC_NAME}"
DetailPrint LANG_ARABIC
${EndSelect}