[tex-live] Problems with non-7bit characters in filename

[TANAKA Takuji]

Hi,

(2014/07/04 17:32), Zdenek Wagner wrote:
> The program has to use a system call to find the filesystem encoding
> and convert the filename from the filesystem encoding to the program's
> internal encoding or vice versa.
I agree that this is the root cause.
In my understanding, the current situation is as follows.

[1] Unix-like (Linux, Mac OS-X) system
[1a] LOCALE of regacy encoding (for example, "de_DE" on linux)
luatex does not work
xetex does not work
due to lack of the filename conversion from the filesystem encoding to the internal encoding (UTF-8)

[1b] LOCALE of utf-8 (for example, "de_DE.UTF-8" on linux)
luatex works
xetex works
since conversion is not needed between the filesystem encoding and the internal encoding (UTF-8)

[2] Windows
[2a] conventional usage
luatex does not work
xetex does not work
due to lack of the filename conversion from the filesystem encoding to the internal encoding (UTF-8)

[2b] set COMMAND_LINE_ENCODING=utf8
luatex does not work (the option is not supported yet)
xetex works
since the option allows to treat Unicode on command line and conversion of filename to the internal encoding (UTF-8)
My implementation in TeX Live 2014 is [2b] for uptex, e-uptex, and xetex.
It is only valid on Windows.

For the case of Klaus-san, there is two options.
[A] Develop to support filename conversion from the filesystem encoding to the internal encoding (UTF-8).
I do not agree since regacy encoding is obsolete.
[B] Switch the file system and LOCALE from regacy encoding to UTF-8
It will work well if the conversion is possible.
I hear "convmv" helps such work.


Takuji Tanaka