Re: ncurses 5.5: Blue Moon demo does not work properly

[Thomas Dickey]

On Thu, 29 Dec 2005, Michael Deutschmann wrote:

> The Blue Moon demo program (program "blue" in the test directory) does
> not render properly on terminals such as "linux" where smpch/rmpch are
> available.
>
> The problem appears to be new code that suppresses the ACS attribute when
> fallback mappings are being used.  This interacts badly with a hack used
> in "blue" -- it assumes that if ??pch is available, it can just set
> ALTCHARSET and plug the raw CP437 values in.

something like that (I'm not sure what "fallback mappings" refers to).
Sorry, I hadn't noticed this quirk was broken, or I'd have fixed it.
That's always only "happened to work", but aside from the "support for PC 
graphic characters" comment in NEWS, was not fully developed.

Some part of that did recently occur to me when I was adding the 
use_legacy_coding() feature for lynx - wondering if I should have
added some feature to support mappings in the C0 area.  I did decide
that I couldn't add that behavior onto the special treatment of C1 area.

> Said CP437 values are ^C, ^D, ^E, and ^F, so when the new code removes
> the ACS attribute, these codes are sent "naked" to the terminal, which
> interprets them as unprintable.  The result is that the columns are
> misaligned, and the suit symbols are absent.
>
> I can workaround the problem by creating and using a modified Linux
> terminfo entry, with "\003\003\004\004\005\005\006\006" added to the acsc
> attribute.

That might be the best solution (except of course for those terminal 
emulators which claim to emulate Linux console ;-).

Or perhaps I could add to the library initialization something that fills 
in those slots in the acs mapping if smpch/rmpch are present and the 
locale isn't UTF-8. (a quick check seems to indicate that this is another 
feature suppressed when the console's in UTF-8 mode).

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