Re: [bug-ncurses] How to use `rep' in xterm since ncurses-6.0-20170729

[Thomas Dickey]

On Tue, Sep 19, 2017 at 06:28:50PM +0200, Dr. Werner Fink wrote:
> On Tue, Sep 19, 2017 at 06:23:16PM +0200, Dr. Werner Fink wrote:
> > Hi
> > 
> > just as subject says and some of ncurses based applications show strange
> > output instead of lines in XTerm  I'd like to ask how to test this with
> > e.g.  tput?
> > 
> 
> Just tried this
> 
>   tput -S <<!
>   > smacs
>   > rep - 20

First, you should update to 20170827:

http://lists.gnu.org/archive/html/bug-ncurses/2017-08/msg00046.html

Next:

tput has to look at the parameters and decide whether they are strings
or numbers.  It has a table to help it assume that certain parameters
are strings.  But a character isn't a string - it fits in better as
a number than as a special case of a string.

The problem is the "-", which isn't handled by tput.  I noticed this and
made a note about a couple of similar cases:

        define_char     defc  ZE Define a character #1, #2 dots wide, descender 
#3
        display_pc_char dispc S1 Display PC character #1

and started to implement another special case for tput's table, but saw
that I was out of time, and put that aside.  Coming back to it, I
decided that the way to work with tput would be to pass the character
value, e.g.,

        rep 45 20

The change that I made at the end of July happened to work for the case
that was reported, but at the end of August, a followup pointed out that
it broke the more general case:

http://lists.gnu.org/archive/html/bug-ncurses/2017-08/threads.html#00039

The current code seems okay now.

As an aside: ncurses won't try to repeat non-ASCII codes, because the
existing logic for UTF-8 vs ASCII line-drawing in ncurses would have to
be integrated with the "rep" (and if you read through PutAttrChar, you'll
see why I didn't do it yet).

-- 
Thomas E. Dickey <address@hidden>
http://invisible-island.net
ftp://ftp.invisible-island.net

signature.asc
Description: Digital signature