Re: quail vs. -nw and the modeline

[Kenichi Handa]

In article <87u1bvlkd1.fsf@jidanni.org>, Dan Jacobson <jidanni@dman.ddts.net> 
writes:

> I do
> $ emacs --no-site-file -q
> then C-\ and in the modeline I see the very nice correct Chinese characters 
> "pin1 fu2"
> However I add -nw
> $ emacs --no-site-file -q -nw
> and type C-\ and instead in the modeline I see jie2, meaning eyelash.
> $ mmencode -q|tr -d = #on the two strings
> ABF7B2C5   B7FB
> Which shows that the string got mangled in the modeline, with BF7B
> extracted and becoming B7FB. 
[...]
> LANG=zh_TW.Big5

Thank you for the report.  I found that, in
leim/quail/pypunct-b5.el, the title of the input-method was
wrongly using GB2312 chars.  It should be Big5 chars.  I've
just installed a fix in RC and HEAD.  Please try the
attached patch.

We need one more fix not to break the terminal by
substituting '?' for unencodable characters on sending it to
the terminal.  Currently that's is done only for iso-2022
based coding systems.  I'll make it work for any coding
systems soon.

---
Ken'ichi HANDA
handa@m17n.org

begin 664 pypunct-b5.el.diff.gz
M'XL(")E0S#X``W!Y<'5N8W0M8C4N96PN9&EF9@"U4%U/PC`4?69_@9?+8F`2
MRL:^-`N2`(D(Q&B4!Q^M78>-8YMMMT#\\[8#C9CXI-Z'V]YSS[GWM/,LIML(
MBEU19D2BIZ!/4^/B]V'<3>\A82F-P":5X'DN;;K!1!QR2MG&?BTQ2^VCW;W*
MX%1R1BN6K8&K0[`\@T$_^`$/C9@E":`2$`'$%;'.H:&V@J"\HCP"DF=)RHC4
M6E%0PA)&L%0#!.0)Y*4L2@E"[E)J=+O=X]]H#$)8E"FXCC-0*7+=R/4!.2H:
MVA9"Z)O`=>$:[[3``\>/M.;\4Q#J#5^CWNCY/=^!N@2PTAS'8.Y_Y_8PV#Q5
M'=VL8133A&44%9B\X#4U6@`F>5:(4-@.?;@QP9P>T,E\ILOFR?C2/WMK6A-3
M#0.IDP(MIQUW9JM5N]=:>,-@VUHJ!N`LAL>J`TG.07-NHH=@M+P::Y[J:T/Z
9]7OSZH;^W;SE>,/1\L_<OP-H]R'5_@(`````
`
end