Re: lynx-dev Big5 charset/attribute problems (was lynx/ncurses log)

[Klaus Weide]

> Hello:
> 
>   Well, just write down what I do, and report it.
> 
> I install Slackware 4.0 in my personal computer.
> 
> 0.
> The software I use:
>   ncurses.tgz --1.9.9e-- bin.tgz in Slackware 3.4
>   ncurses.tgz --1.9.9g-- bin.tgz in Slackware 4.0
>   ncurses-1.9.9g.tgz -- source code
>   ncurses-5.0-990724.tgz -- source code
>   lynx.tgz --lynx2.8.1rel2-- bin.tgz in Slackware 4.0
>   lynx2-8-3dev.4.tar.gz -- source code
>   lynx2-8-3dev.5.tar.bz2 -- source code
> 
> The configuration options i use for source code:
>   ncurses-1.9.9g: the same as below
>   ncurses-5.0-990724: prefix=/usr/local exec-prefix=/usr/local
>     with-shared with-normal with-debug with-profile
>     enable-symlinks enable-overwrite
>   lynx2-8-3dev.4: prefix=/usr/local libdir=/usr/local/lib/lynx
>     enable-default-colors with-screen=ncurses enable-gzip-help
>   lynx2-8-3dev.5: the same as above and test with
>     enable/disable-cjk
> and before configure lynx, follow the instructions of INSTALLATION 1d.
> to set LIBS="-L/usr/local/lib" and CPPFLAGS="-I/usr/local/include"
> 
> 1. Situation A:
> ncurses-1.9.9g.tgz / ncurses-5.0-990724.tgz + lynx2-8-3dev.4.tar.gz
> ( Of course, I uninstall one ncurses before I install the other one )
> 
>   Well, ncurses-1.9.9g really cannot work correctly with Chinese word
> set, but it is ok, since we have ncurses-5.0-beta version. :)
> 
>   I focus on Chinese character set only:
> Assumed document character set:
>   1a. big5
>   1b. iso-8859-1
> Display character set:
>   1a. iso-8859-1
>   1b. Chinese or Taipei(Big5)
>   Well, when both char-set 1a. some Chinese word breaks(double-byte), but
> hyperlink/directory's color(foreward/background) works as lynx.cfg set.
> Though i export TERM=vt100 hyperlink/directory works ok(foreware=
> lightwhite/background=black) too, and normal text is normalwhite/black.
> ( That's why I like work with TERM=vt100 lynx. Long period reading won't
> hurt my eyes too much :).
>   Ah... if both char-set 1b. All Chinese words no break and show correctly.
> BUT when TERM=vt100, all normal text and hypertext/directory all normal 
> white/black.
> After arrow-key moves, hypertext/directory become lightwhite/black, but after
> ctrl-l all become normalwhite/black again. ( Oops... I forget test TERM=linux,
> color mode. )
>   Assumed char-set: iso-8859-1 and Display char-set: iso-8859-1 works as 1a.
>   Assumed char-set: big5 and Display char-set: Chinese or Taipei(Big5) works 
> as 1b.

In other words, behavior depends only on Display character set, not 
assumed charset.

So with 'Display character set' set correctly you get the right display,
except for attributes.

> Well, how can I get 1. no break Chinese Word and 2. TERM=vt100 normaltext 
> color
> normalwhite/black and hypertext/directory color lightwhite/black at the same
> time ? lynx.cfg so large... :)

There was something that just came up in connection with UTF-8.  I don't
know whether it has something to do with it, but please try whether
it helps you:  Visit the list archive for this month (July) under
<http://www.flora.org/lynx-dev/html/> and read the messages with
"character-set=utf-8" in the subject.  Try taking out the lines
of code suggested there.  Also try just replacing the
'lynx_force_repaint();' with 'clearok(curscr,TRUE);'.

Also try the 'strace' I suggested somewhere in that thread.


> 2. Situation B:
> ncurses-1.9.9g.tgz / ncurses-5.0-990724.tgz + lynx2-8-3dev."5".tar.gz
> 
>   Well, no matter what ncurses version it is, and whether enable/disable-cjk.
> I follow the configuration above...
> 
>   No matter what assumed/display character set I try, all text, hypertext, dir
> ectory foreward is normal-white, background is black. ( Oops again, I forget
> TERM=linux situation... )
>   AND no matter what char-set I try, Some Chinese Words cannot be recongized.
> Not double byte cut/break, cannot be recongized. I don't know what word it is,
> not Chinese Word.

I guess what you mean is that characters are replaced by (wrong)
7-bit ASCII equivalents?

Maybe something broke for CJK between dev.4 and dev.5.

But are you sure you were testing exactly the same conditions as for
dev.4?

Even though the document's (assumed) charset doesn't seem to play a role,
I would try to label the document you are testing with correctly as to
charset.
Put <META http-equiv="content-type" content="text/html;charset=big5">
in the <HEAD> section of a local copy.  Or better, use a HTTP-served
document that is sent with an explicit HTTP "Content-Type" header that
includes "charset=big5", if possible (see ']' HEAD command for checking
HTTP headers).

Try to always start a new instance of lynx after changing a character
set setting, to avoid any bugs with using the previous setting directly
after a change. (I'm not saying there are such bugs, but just to make
sure.)

> 3.
>   ncurses-1.9.9g no need discuss, it conflicts with lynx Chinese set without 
> saying.

(I don't know how.)

> Oh, the Chinese Terminal/Console program I use: yactp4.tar.gz
> It based on svgalib 1024x768x256 and 24x24 fonts shows. 
> Chdrv only support 640x480 15x16 fonts.
> 
> Sorry again, poor english... :p
> -- 
> address@hidden
> 
>