From Newsgroup: comp.unix.shell

Kornshell doesn't seem to handle umlauts or other non-ASCII Unicode
characters correctly with the 'select' statement; the display shows
(for example)

1) abcdefghijklmnopqrstuvwxyz 15) abcdefghijklmnopqrstuvwxyz
2) abcdefghijklmnopqrstuvwxyz 16) abcdefghijklmnopqrstuvwxyz
3) |nbcdefghijklmnopqrstuvwxyz 17) ABCDEFGHIJKLMNOPQRSTUVWXYZ
4) |nbcdefghijklmn||pqrstuvwxyz 18) ABCDEFGHIJKLMNOPQRSTUVWXYZ
5) |nbcdefghijklmn||pqrst|+vwxyz 19) |aBCDEFGHIJKLMNOPQRSTUVWXYZ
6) abcdre4fghijkl-|nopqrstuvwxyz 20) |aBCDEFGHIJKLMN|uPQRSTUVWXYZ
7) abcdre4fghijklmnopqrstuvwxyz 21) |aBCDEFGHIJKLMN|uPQRST|LVWXYZ
8) |nbcdefghijklmn||pqr|ft|+vwxyz 22) ABCDEFGHIJKLMNOPQRSTUVWXYZ
9) abcdefghijklmnopqrstuvwxyz 23) ABCDEFGHIJKLMNOPQRSTUVWXYZ
10) abcdefghijklmnopqrstuvwxyz 24) |aBCDEFGHIJKLMN|uPQR?T|LVWXYZ
11) abcdefghijklmnopqrstuvwxyz 25) ABCDEFGHIJKLMNOPQRSTUVWXYZ
12) abcdefghijklmnopqrstuvwxyz 26) ABCDEFGHIJKLMNOPQRSTUVWXYZ
13) abcdefghijklmnopqrstuvwxyz 27) ABCDEFGHIJKLMNOPQRSTUVWXYZ
14) abcdefghijklmnopqrstuvwxyz

Somehow it obviously gets confused with the count of the number of
characters and the count of the octets in the encoding, thus the
formatting gets corrupted.

(Playing with other locales doesn't change that effect.)

Observed in ksh version AJM 93u+m/1.0.8 2024-01-01.

(Bash handles that correctly.)

Janis

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

Op 11-01-2026 om 04:51 schreef Janis Papanagnou:

Kornshell-adoesn't-aseem-ato-ahandle-aumlauts-aor-aother-anon-ASCII-aUnicode characters-acorrectly-awith-athe-a'select'-astatement;

The code for showing the menu is not aware of multibyte locales. Thanks for the report. I've fixed it for the next release. See: https://github.com/ksh93/ksh/commit/18e4cbc1

Observed in ksh version AJM 93u+m/1.0.8 2024-01-01.

FYI, you're two point releases behind.
--
|| modernish -- harness the shell
|| https://github.com/modernish/modernish
||
|| KornShell lives!
|| https://github.com/ksh93/ksh
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

Thanks for fixing the select issue!

On 2026-02-25 01:39, Martijn Dekker wrote:

Op 11-01-2026 om 04:51 schreef Janis Papanagnou:

[...]
Observed in ksh version AJM 93u+m/1.0.8 2024-01-01.

FYI, you're two point releases behind.

Thanks. - That's probably because since my distro meanwhile supports
"u+m" already I'm just using what's provided.[*]

But maybe I should again get the newest one and ignore what comes out
of the box with my system? (I think if any of the two other issues I
have with "u+m" would get fixed then I'd do that, but I recall you
haven't intended to change these[**], so there's [at the moment] no
pressing need to change to the latest release.)

Janis

[*] If it were still an "u+" I'd certainly get "u+m" from a more up to
date source since I prefer that. I'm glad that "u+m" made it into the
distros.

[**] One was the vi-mode expansion of $(cmd) when typing a '*', which
was already _existing_ in original "u+" (and which, sadly, you removed
from "u+m"); I had used that very often and I'm really missing it. :-/

--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

On 25/02/2026 00:39, Martijn Dekker wrote:

Op 11-01-2026 om 04:51 schreef Janis Papanagnou:

Kornshell-adoesn't-aseem-ato-ahandle-aumlauts-aor-aother-anon-ASCII-aUnicode >> characters-acorrectly-awith-athe-a'select'-astatement;

The code for showing the menu is not aware of multibyte locales. Thanks
for the report. I've fixed it for the next release. See: https://github.com/ksh93/ksh/commit/18e4cbc1

Observed in ksh version AJM 93u+m/1.0.8 2024-01-01.

FYI, you're two point releases behind.

Same for printf "%-*s" - counts octets, not printable characters.

93u+m/1.0.6, so way behind.



--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

Richard Harnden wrote:

Same for printf "%-*s" - counts octets, not printable characters.

That's what printf is supposed to do, as POSIX requires that field
widths specify the number of bytes (I assume for compatibility with
the printf() C function).
--
Geoff Clare <netnews@gclare.org.uk>
--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

On 17/03/2026 13:31, Geoff Clare wrote:

Richard Harnden wrote:

Same for printf "%-*s" - counts octets, not printable characters.

That's what printf is supposed to do, as POSIX requires that field
widths specify the number of bytes (I assume for compatibility with
the printf() C function).

Ah, okay, thanks.

I have a function that pads by ignoring utf8 continuation bytes and
ignores ansi escapes. Not exactly pretty, but it works for me.

--- Synchronet 3.21d-Linux NewsLink 1.2