Re: [Lynx-dev] Unicode display problem

[Thomas Dickey]

On Mon, 28 Jun 2004 address@hidden wrote:

> 040628 Thomas Dickey wrote:
> > On Mon, Jun 28, 2004 at 03:06:04PM -0400, address@hidden wrote:
> >> i've been testing how far i can use Unicode for Ancient Greek & Esperanto.
> >> one thing i've found is that -- following instructions --
> >> only some of the characters appear with Lynx.  my test file is attached.
> >> on Gentoo Linux + Lynx 2.8.5dev16, i created a virtual terminal via
> > of course 2.8.5dev.16 is rather old (more than a year).
> > But the underlying version of ncurses is more important.
> > If it's not the wide-character configuration of ncurses,
> > your results won't be as good.
>
> all Gentoo has to offer is :
>
>   sys-libs/ncurses
>       Latest version available: 5.4-r1
>       Latest version installed: 5.4-r1
>       Size of downloaded files: 2,103 kB
>       Homepage:    http://www.gnu.org/software/ncurses/ncurses.html
>       Description: Linux console display library

Well, as I suggested in my previous message, there's nothing that I know
that would break if you installed the ncursesw libraries.  (At some point
Gentoo will provide the package, though I've seen some comments that
they're not happy that the two flavors have different names ;-).

>
> >> further, in the test file obtained via Lynx help (Options menu),
> > are you talking about setting display character set to something?
>
> no: if you do  'o' [3] [9] [36] , you get a test file.

oh (I hadn't noticed that file, since the ones in keystrokes that I
_look_ at are the HTML files).  I occasionally look at the test cases, but
directly from the build tree.  There are several interesting ones (in
the test directory of lynx's source):

ALT88592.html
ISO_LATIN1_test.html
README.txt
TestComment.html
c1.html
iso8859-1.html
iso88592.html
koi8-r.html
quickbrown.html
raw8bit.html
sgml.html
spaces.html
tabtest.html
unicode.html
utf-8-demo.html

> >> i can see only  00A9 03BB 042E 043B , but none of the others.
> > Fonts can be a problem.  Checking, I see the characters using
> > *VT100*font5:       
> > -misc-fixed-medium-r-normal--18-120-100-100-c-90-iso10646-1
>
> that's what i specified (at least, the part after the ':').
>
> > *VT100*font6:       
> > -misc-fixed-medium-r-normal--20-200-75-75-c-100-iso10646-1
> > While the latter is more readable, I've found that it isn't as complete.
>
> same effect.  do you see all  24  Greek letters in my own test file ?

yes - I considered making a picture (can do that if you're curious).

> > But for this example, either works (for me - using ncursesw).
>
> i can't find anything like 'ncursesw' on my system.
>
> > Perhaps one of these:
> >     a) using slang or ncurses rather than ncursesw

no - slang does no better than ncurses.  There's a slang-utf8 floating
about, but slang's author refuses to touch it.

> >     b) using an older version (?) of the fonts
> >     c) using an older version of ncursesw
> >        5.4 is recommended - best results with the rollup patch from April

5.3 was the first that I had usable UTF-8.  There were (inevitably) other
things found after that, which I fixed in 5.4.  I'm mainly working on
secondary things (things that broke as a result of a fix or weren't good
enough).  5.4 with the most recent rollup patch (see my ncurses faq) in

        ftp://invisible-island.net/ncurses/5.4/

is the best I can recommend.

>
> below is my configure command (edited to split lines):
>
>   ./configure --with-screen=ncurses --disable-menu-options
                --with-screen=ncursesw

-- 
Thomas E. Dickey
http://invisible-island.net
ftp://invisible-island.net