[BRLTTY] can't read screen content

[Samuel Thibault]

Aura Kelloniemi, le ven. 22 mai 2026 12:19:27 +0300, a ecrit:
> On 2026-05-22 at 09:53 +0200, Samuel Thibault <samuel.thibault at ens-lyon.org> wrote:
>  > Sébastien Hinderer, le ven. 22 mai 2026 07:19:59 +0200, a ecrit:
>  > > By the way: am I correct that BRLTTY has no way to know dynamically
>  > > whether the console is in UTF-8 mode or not?
> 
>  > It is IUTF8 in the termios c_iflags, but that should only be relevant
>  > for console input, not for output, that shouldn't be changing the way
>  > /dev/vcsa reports characters, so that's most probably just a bug in the
>  > kernel.
> 
> I would think this is so that /dev/vcsa works all the time (because switching
> BRLTTY's vcsu support off makes it work), but /dev/vcsu devices stop working
> once the terminal is in non-UTF-8 mode.
> 
> It is hard to imagine, what /dev/vcsu devices should present, when the console
> contents are not Unicode at all.

I don't see a reason to stop /dev/vcsu support when not in UTF-8 mode.
The kernel knows what unicode characters we have, it can produce
/dev/vcsu.

> However, the problem still remains:how could BRLTTY know which
> character set is used by the console, if it is not UTF-8.

This is the font map, that brltty already knows to fetch with
GIO_UNIMAP.

Samuel