tlpkg UTF-8 Umlauts Problem

[Siep Kroonenberg]

On Wed, Mar 24, 2021 at 11:21:10AM +0100, Hartmut Henkel wrote:
> Dear Siep,
> 
> On Wed, 24 Mar 2021, Siep Kroonenberg wrote:
> > On Wed, Mar 24, 2021 at 09:56:23AM +0100, Hartmut Henkel wrote:
> > > when TeX Live is installed, it shows (in german):
> > >
> > >   Schauen Sie in die /usr/local/texlive/2021/index.html fr Links auf die Dokumentation.
> > >   Die TeX Live Website (https://tug.org/texlive/) enthlt Updates and
> > >   Korrekturen. TeX Live ist ein vereintes Projekt der TeX Nutzergruppen
> > >   weltweit, werden Sie Mitglied um ihre Arbeit zu untersttzen. Die Liste
> > >   der Gruppen finden Sie unter https://tug.org/usergroups.html.
> > >
> > >   Fge /usr/local/texlive/2021/texmf-dist/doc/man zu  MANPATH hinzu.
> > >   Fge /usr/local/texlive/2021/texmf-dist/doc/info zu INFOPATH hinzu.
> > >
> > > Actually the message above shows in the gnome term (linux) all
> > > latin-small u with diaeresis (UTF-8 C3 BC) as "replacement character"
> > > (UTF-8 EF BF BD). (This got lost when copying into the mail.)
> > >
> > > With the same terminal i can read de.po without problem, and all umlauts
> > > are there. Hexdump shows that in de.po they are proper 2-byte UTF-8. So
> > > at which point between de.po and the Tex Live installer message on the
> > > terminal these umlauts might get mangled?
> >
> > With urxvt I see the umlauts just fine. What do you get with another
> > terminal? And with the gui installer?
> 
> (always fresh install, with dir. 2021/ removed before)
> 
> The "install-tl -gui" installer does not translate here, it shows all in
> english (?!).

The GUI relies on Tcl/Tk's msgcat module to determine the default
language if no language is specified. I believe that the LC_MESSAGES
environment variable is the one that matters:

LC_MESSAGES=de_DE.UTF-8

Anyhow, the menu bar lets you switch languages on the fly.

> > According to gucharmap is C3 BC in the Hangul Syllables unicode block.
> 
> According to an UTF-8 table this is ü / c3 bc / LATIN SMALL LETTER U
> WITH DIAERESIS. And similar install-tl problem with ä.

Gucharmap tells me: U+00FC LATIN SMALL LETTER U WITH DIAERESIS.

On an Ubuntu Gnome test system I could reproduce the problem,
although here gucharmap gave the same 00FC code as on my regular
system (archlinux/fvwm).

Odd.

-- 
Siep Kroonenberg