Closed ryzom-pipeline closed 8 years ago
Original comment by Cédric Ochs (Bitbucket: [Cédric OCHS](https://bitbucket.org/Cédric OCHS), ).
Yep, and it'll be a huge task since fopen is used almost everywhere :p
Original comment by Jan Boon (Bitbucket: [Jan Boon](https://bitbucket.org/Jan Boon), ).
Make a NLMISC::fopen that converts from utf8 to wchar on Windows and uses the wchar fopen variant there
Original comment by Cédric Ochs (Bitbucket: [Cédric OCHS](https://bitbucket.org/Cédric OCHS), ).
Yes, I thought about that this night :)
Original comment by Cédric Ochs (Bitbucket: [Cédric OCHS](https://bitbucket.org/Cédric OCHS), ).
I need to work on it for next major version, because some users can have a username with unicode characters and since Ryzom will be installed in their Local profile, NeL could have some problems loading these files.
I already did some checks and even if using only "8 bits locale" everywhere works for French (é character for example), it'll fail for more complex characters like Cyrillic or Asian ones.
We have only 2 choices :
Original comment by Jan Boon (Bitbucket: [Jan Boon](https://bitbucket.org/Jan Boon), ).
Yeah, it's best to consistently use UTF-8 everywhere, since resources like text files are also being loaded as UTF-8 and if they contain file names they'll technically be UTF-8 format anyway.
Original comment by Cédric Ochs (Bitbucket: [Cédric OCHS](https://bitbucket.org/Cédric OCHS), ).
Yep I agree :)
Original comment by Cédric Ochs (Bitbucket: [Cédric OCHS](https://bitbucket.org/Cédric OCHS), ).
Last changes for Unicode support under Windows, fixes #261
Original report by Jan Boon (Bitbucket: [Jan Boon](https://bitbucket.org/Jan Boon), ).
Use the WCHAR version of all OS functions that deal with paths, and convert to UTF-8 std::string.