Bug#254316: CJK font fails after running slang/ncurses program

[Juliusz Chroboczek <jch@pps.jussieu.fr>]

Hi Thomas!

>> The latest xterm seems to be able to display CJK characters.
>> However, I found that after running a slang / ncurses application,
>> the characters become garbled again.

Geoffrey: as a workaround, you may do

  TERM=kterm; export TERM

> So, if that's true, does that mean it is luit related and not xterm
> related?

It's a bug in terminfo that I've alreay reported to you, Thomas, when
I first wrote luit.  In the xterm cap, there's the following:

  enacs=\E(\B\E)0
  smacs=^N
  rmacs=^O

This clobbers G1.  It works in ISO 8859-n locales because there's a
workaround in luit (luit uses G2 instead of G1 for such locales).
While there's in principle nothing preventing the same workaround to
work for Big5 locales, the problem with EUC will remain.

The proper fix is to use the following in the xterm cap:

  enacs=
  rmacs=\E(B
  smacs=\E(0

This only uses G0, and works in any locale that has ASCII in G0, which
includes all standard Unix locales.

Branden: please reassign this to ncurses-base and mark it upstream.

                                        Juliusz