TEXMFHOME on Windows (for users with long names, diacritics or spaces in their names)

[ttk at t-lab.opal.ne.jp]

Hi John, all,

I intend to support non-ASCII Unicode characters in pdf(la)tex on Windows since TeX Live svn r47973 (Jun 2018).
It runs when we set kpathseach parameter "command_line_encoding=utf-8".
I think "command_line_encoding=utf-8" is current default setting.

On 2024/05/25 9:22, John Collins wrote:
> 4. When I ran 'kpsewhich uuuu.tex', it reported successfully finding the file, but with the apparently wrong name for the directory.

How about `kpsewhich -progname=pdftex uuuu.tex` ?

> 5. I used pdflatex to compile a .tex file that had the line  \input{uuuu} in it.  The file uuuu.tex was correctly read, and pdflatex reported the correct directory name.

I wonder whether the "a .tex file" is encoded in UTF-8 or in CP1252 or in another 8-bit encoding.
If the "a .tex file" is encoded in UTF-8, the behavior is what I intended.

> Further tests showed that kpsewhich appears to be coding its output in the system code page, which on my system (and probably a default Windows system in France) is CP1252.  Simply doing 'chcp 1252' resulted in a correct display from kpsewhich.

If the "a .tex file" is encoded in UTF-8 and "command_line_encoding=utf-8" is set,
`chcp 65001` and `kpsewhich -progname=pdftex uuuu.tex` will work.
I expect it will work both on cmd.exe and power shell.


Thanks,
Takuji