Re: [Freedos-devel] country.sys changes WAS: Re: release kernel 2044

[Rugxulo <ru...@gm...>]

Saluton, amiko,

On Mon, Jan 26, 2026 at 10:53 PM perditionc--- via Freedos-devel
<fre...@li...> wrote:
>
> These changes need some additional discussion and will ultimately need
> someone that understands the languages to review.
>
> See https://github.com/FDOS/kernel/issues/235 and corresponding pull
> request in country Github source referenced.
>
> Specific concerns:
> esperanto has no country, I don't think it can be added.

"In 1954, the United Nations — through UNESCO — granted official
support to Esperanto as an international auxiliary language in the
Montevideo Resolution. However, Esperanto is not one of the six
official languages of the UN."

I'm not aware of any active countries that have E-o officially recognized.
(Zamenhof was Polish, so using Poland as country wouldn't be horribly
incorrect.)

* https://en.wikipedia.org/wiki/Esperantujo

> My understanding is that one would pick a country with a compatible codepage to use it.

FreeDOS supports E-o with CP 853 (with KEYB and DISPLAY support), e.g.
EGA.CPX. Compared to the traditional ISO-8859-3 (aka Latin-3), it also
supports box chars for text-mode windows. But the two codepages are
incompatible (although they use the same letters but in different
spots). In other words, AFAIK, without COUNTRY.SYS, you can't switch
"system" codepage, e.g. "chcp 853". (I would guess PRINTER.SYS is
irrelevant here.) COUNTRY.SYS also affects NLSFUNC, right?

Latin-3, aka CP 913, is more popular, loosely speaking, but even that
is somewhat considered obsolete. Some OS/2 versions probably supported
it, not fully sure. FreeDOS does not support this (last I checked).
Kosta Kostis had some freeware tools (with sources) on his website
(e.g. ISOLATIN.CPI) that supported it. I also know DR-DOS 7.03 (1999),
EGA.CPI, had undocumented Latin-3 support too (but was missing
"lowercase j with circumflex") but no KEYB support. There was also a
small third-party ILOJ.TGZ (TSR from early '90s with .ASM source) that
allowed Latin-3 display (EGA+) and input.

I did submit a patch to Mined, years ago, that allowed for better
DISPLAY checking (FreeDOS or MS-DOS or DR-DOS). GNU Emacs used to have
its tutorial in E-o via Latin-3, but obviously Unicode is preferred
these days. (We did get a new DJGPP build a few months ago.)

Realistically, it might just be easier to write in 7-bit ASCII (e.g.
cx or c^ for "lowercase c with circumflex" or ux or u( for "uppercase
U with breve") and manually translate it to UTF-8 with a simple tool.
The "x method" was very popular in news://soc.culture.esperanto and
other online places, but "officially" you just use "ch" for "c^" etc.
(and "u breve" just becomes "u") via the "h method". In short, UTF-8
still uses two bytes per accented character, so it's not really
"better" than old-fashioned (x or h) methods or even cp853 or cp913,
at least in storage space.

I don't do a lot of E-o reading anymore (and am not a paying member of
E-USA), but I still have a small shelf full of books and periodicals.