Re: wide characters, get_wch() and form_driver() - cannot get them working together.

[Thomas Dickey]

On Wed, 18 Jun 2008, Adam Spragg wrote:

> Hi,
>
> On Tuesday 17 June 2008 14:57:01 Thomas Dickey wrote:
> > The resolution was basically saying that the form library gets
> > compiled to support wide-character mode, and that _it_ knows only
> > about byte-at-a-time adds via addch, but that (with the limitation of
> > not moving the cursor in the middle of the operation), should work.
>
> I've compiled my own ncurses library from the gnu sources, rebuilt the
> test program against that, linked with it, run it and got the same
> problem - E_UNKNOWN_COMMAND when adding the first character of a '£',
> random accented vowels from unicode code points 192-255, and '€' - the
> euro symbol. When trying to enter the Euro symbol, the status line
> prints out the correct character id of 8364 (0x20ac) and points out
> that I got the E_UNKNOWN_COMMAND on byte 0 of 3, which is the correct
> number of bytes for the utf-8 encoding of that character.
>
> I configured ncurses with the command line:
>  ./configure --prefix=/home/adam --enable-widec --with-shared

I didn't get that far (was doing maintenance for $dayjob last night).

> and checked the test program was loading the new copy of the library
> with strace.
>
> Anything else I can check on my end?

Aside from perhaps suspecting the chunk at the end of form_driver
(which looks okay at the moment):

      /*
       * If we're using 8-bit characters, iscntrl+isprint cover the whole set.
       * But with multibyte characters, there is a third possibility, i.e.,
       * parts of characters that build up into printable characters which are
       * not considered printable.
       *
       * FIXME: the wide-character branch should also use Check_Char().
       */
#if USE_WIDEC_SUPPORT
      if (!iscntrl(UChar(c)))
#else
      if (isprint(UChar(c)) &&
          Check_Char(form->current->type, c,
                     (TypeArgument *)(form->current->arg)))
#endif
        res = Data_Entry(form, c);
    }

For investigating your core dump, I intended to build a debug-version,
and run it with valgrind (perhaps this evening).

> Adam
>
> p.s. Just out of curiosity, are you seeing the '£' (pound) and '€'
> (euro) characters correctly in the emails I'm sending? I've checked my
> outgoing mail and as far as I can tell it's being sent in utf-8 and is
> identifying itself as utf-8. But your earlier reply to had the charset
> set to X-UNKNOWN which caused some problems displaying the pound
> character, and the archived version at lists.gnu.org[0] isn't quite
> right either.

at the moment, I'm connected to my mail-provider in pine, and that shows
the Latin-1 codes (no UTF-8).  On my home machine, that would work 
properly.

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