Re: Question: How does ncurses store and handle "wide" characters?

[Thomas Dickey]

On Thu, Jun 03, 2021 at 01:11:08PM -0400, Bill Gray wrote:
> Hi Peter,  all,
> 
>    Just a note : the PDCursesMod documentation is almost
> entirely from the contributors to PDCurses,  not me.  The
> bit about "standing on the shoulders of giants" applies.
> 
>    In this case,  you really _are_ reading the C headers;
> it's just that MANUAL.md is extracted from them (and the
> C source).  I hadn't really thought about it much (it was
> that way when I got here),  but I think that when I'm trying
> to look something up in PDCurses/PDCursesMod,  I usually
> look in the headers or source.  However,  there have been
> times when just searching MANUAL.md made more sense.
> 
>    I should also note that I have a foot on Thomas' shoulders
> as well.  My knowledge of Curses standards is quite minimal.
> I mostly just figure that if ncurses does something in a
> particular way,  it'll either be the standard way or will
> be documented as "warning,  this is non-standard",  and
> it'll be the way most people are expecting it to work.
> 
>    At this point,  given the "market share" of ncurses,
> it's become a de facto standard.  When PDCursesMod doesn't
> behave as ncurses does,  I generally think there ought to
> be a good reason.  Sometimes,  there is (as with PDCurses'
> implementation of full RGB colors).  Usually not,  though.
> 
>    Which leads me to ask,  Thomas,  about this comment :
> 
>    "...actually you're telling me that PDCurses has an
> unnecessary deviation from X/Open Curses."

I was misreading his comment about NCURSES_WIDECHAR.
Actually he was referring to the comment-header in PDCurses
which tries to show the symbols that a user might want to use to
turn on features.

Looking at the 3.4 headers (on Linux...), PDCurses has this:

    XCURSES         True if compiling for X11.
    PDC_RGB         True if you want to use RGB color definitions
                    (Red = 1, Green = 2, Blue = 4) instead of BGR.
    PDC_WIDE        True if building wide-character support.
    PDC_DLL_BUILD   True if building a Win32 DLL.
    NCURSES_MOUSE_VERSION   Use the ncurses mouse API instead
                            of PDCurses' traditional mouse API.

but the "XCURSES" is misplaced (it should be set by the implementation,
and PDC_DLL_BUILD is irrelevant on Linux).

The others are set by the user.

In ncurses, there's no summary of set-by-the-user.  ncurses' curses.h is
more than twice as long as that for PDcurses, with 3-4 times as many "#if"
lines.  But most of that is for constants (settings done in the configure
script). Offhand, there's about as many symbols that the user might
define -- which might be better expressed in the manpage, since a one-line
comment doesn't help that much, to explain why a feature is compile-time
configurable.

Looking quickly, I see

        NCURSES_ATTR_T (normally not set by users)
        NCURSES_INTERNALS (normally not set by users)
        NCURSES_WATTR_MACROS
        NCURSES_REENTRANT
        NCURSES_WIDECHAR
        NCURSES_NOMACROS

Most of the other symbols are set by the configuration scripts.

(I'm reminded that I should get rid of "TRACE", which is very old,
e.g., replacing it with "NCURSES_TRACE",
but can't simply drop things that others might be using...)

Following up on the bit-layout, though: I use the symbols all the time,
and don't look at the hex/octal values of the mouse-masks or attributes.

>    I'm a little lost here.  I gather you're thinking the
> wide-character support in PDCurses is enabled by default?
> It isn't;  you have to compile with PDC_WIDE enabled.  Omit
> that,  and only eight of the 21 character bits actually
> get used.  Or are you seeing some other deviation?
> 
> Thanks!    -- Bill

-- 
Thomas E. Dickey <dickey@invisible-island.net>
https://invisible-island.net
ftp://ftp.invisible-island.net

signature.asc
Description: PGP signature