Re: [vile] I've got my 'wide' characters display problem back - in 9.7z

[Chris G]

On Thu, Jan 28, 2010 at 07:06:18AM -0500, Thomas Dickey wrote:
> >>
> >>>This time I *have* compiled with:-
> >>>
> >>>  ./configure --with-perl --with-screen=ncursesw
> >>
> >>It seems to work for me, here.  I have a set of test-files that I
> >>use for checking the Latin-1 codes versus the Unicode values which
> >>are the "same"
> >>
> >>Here
> >>
> >>:show-variables
> >>
> >>shows (among other things):
> >>
> >>$buf-encoding = UTF-8
> >>$cfgopts = hypertext,locale,iconv,multibyte,perl,curses
> >>
> >Yes, I have:-
> >   $buf-encoding = UTF-8
> >   $cfgopts = hypertext,locale,iconv,multibyte,curses
> >
> >I'm not quite sure why I haven't got perl but that's presumably
> >irrelevant to the present problem.
> 
> That's because I configure --with-perl so that I can spot build-issues,
> etc.
> 
See above, so did I!  :-)

> >
> >
> >>Also, in :setall, I have nounicode-as-hex (near the bottom, in the
> >>window section).
> >>
> >Yes, I have that too.
> >
> >>>
> >>>Locale settings are all en_GB.UTF-8, platform is xubuntu 9.10 with
> >>>2.6.31-17-generic kernel.
> >>>
> >>>The characters are actually correct, it's just their display in vile
> >>>that's wrong.
> >>>
> >
> >Anything else I can try/do to diagnose it?
> 
> ldd vile should of course show libncursesw (I assumed that...).
> 
Yes:-
    chris$ ldd /usr/local/bin/vile
            linux-vdso.so.1 =>  (0x00007fff30bff000)
            libcrypt.so.1 => /lib/libcrypt.so.1 (0x00007f9b1d487000)
            libncursesw.so.5 => /lib/libncursesw.so.5 (0x00007f9b1d238000)
            libc.so.6 => /lib/libc.so.6 (0x00007f9b1cec9000)
            libdl.so.2 => /lib/libdl.so.2 (0x00007f9b1ccc5000)
            /lib64/ld-linux-x86-64.so.2 (0x00007f9b1d6c0000)

> Beyond that - I diagnose most of the problems by configuring --with-trace,
> and looking at the Trace.out file when running it.  That's large, unless
> one uses an empty .vilerc file.  (If the problem is in the .vilerc file,
> then I pare that down to just the problem before wading through 4Mb
> of traces...).
> 
I've removed my .vilerc file and the problem is still there.  I'll
configure with --with-trace and take a look.

> The basic locale initialization is near the beginning, but the decision of
> whether the buffer should be utf8 (probably not the problem) is at
> the end.
> 
> -- 
> Thomas E. Dickey
> http://invisible-island.net
> ftp://invisible-island.net

-- 
Chris Green