Re: lynx-dev Display of SGML Greek Math entities -- solution

[Steve White]

Regarding the display of euc-jp encoded pages with lynx.

I can't get lynx to work right.  Please explain.

Henry provided a page showing the Greek alphabetic characters encoded
as raw euc-jp and as numeric character references.

He also provided image files of his display.  Corresponding raw euc-jp
and numeric character references are displayed identically.  He's
using TeraTerm on a Windows system communicating with a unix machine
(I think...).

I have never managed to get lynx to display both correctly on my
system (Linux Mandrake 8.2). 

I've tried xterm, kterm, and rxvt.  I think the correct setup to
display multibyte character sets for each terminal is as follows:

        kterm   runtime argument   -km euc
        rxvt    compile define     KANJI
        xterm   runtime argument   -u8

Lynx should be run with the runtime argument that matches the display
capability of the terminal: -display_charset=euc-jp or utf-8.

The result is this: 

  * In the euc-jp enabled terminals rxvd and kterm, raw euc-jp
    characters display as Greek, but numeric character references
    are transliterated into Latin characters.

  * In the Unicode-enabled xterm, raw euc-jp characters are junk,
    while numeric character references display as Greek.

It is pretty clear that lynx is doing this.  Neither behavior is right.

Could it be a compilation option?  I tried compiling lynx with
KANJI_CODE_OVERRIDE defined in userdefs.h.  Also messed with
USE_TH_JP_AUTO_DETECT and CONV_JISX0201KANA_JISX0208KANA, to no
good effect.

Could lynx be trying to detect the display capabilities of the
terminal internally?

On Mon, 19 Aug 2002 13:29:37 -0400 (EDT)
Jungshik Shin <address@hidden> wrote
>
>  Under X, UTF-8 is certainly the best way. Anyone can go to
>Thomas's page and grab the newest xterm-16x (http://dickey.his.com/xterm).
>If Unix one's using doesn't support UTF-8 locale, just run it with
>'-u8' option with some free iso10646-1 X11 fonts.  display-charset of
>Lynx can be set to utf-8 and it'll work perfectly for web pages encoded
>in single byte legacy encodings as well as in UTF-8. For MS-Windows,
>there are a couple of pretty good terminal emulators that support UTF-8
>(putty is one of them) under which one can connect to a remote Unix host
>(ssh and telnet if somebody still wants to use it). However, as I wrote in
>my message Henry refered to, this does not yet work for CJK (multibyte)
>legacy encodings because Lynx does not know how to convert between
>Unicode/10646 and CJK encodings. That's why '&alpha;' is rendered as
>'a' when display charset is one of CJK encodings (i.e. the same way as
>it's rendered under ISO-8859-x terminal where x is not 7) although CJK
>legacy encodings CAN represent  'alpha' (albeit in double-width).
>



; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden