Re: mvwprintw

[Bob Rossi]

On Fri, Mar 31, 2006 at 03:26:55PM -0500, Bob Rossi wrote:
> On Fri, Mar 31, 2006 at 02:53:44PM -0500, Thomas Dickey wrote:
> > On Fri, 31 Mar 2006, Bob Rossi wrote:
> > 
> > >OK, here is the script file that has the error. I also have the the
> > >output of unmap attached. I am having a little trouble reading these
> > >files for the first time. I suppose I better get good at it.
> > >
> > >See anything interesting?
> > 
> > I don't see the \E[1C split-up in this trace, but find it unusual to
> > see a long string of nulls (^@) after the chunk:
> > 
> > \n^N^O~^Nx^O
> > \E[27;4H^N^O
> > \E[60;8Hb main.c:777777777777
> > \E[21Db 
> > main1Cc:address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@address@hidden@^@
> > 
> > There's not much here to read:
> >     ^N and ^O switch to/from line-drawing mode
> >     \E[60;8H moves the cursor to cell 60,8
> >     \E[21D moves the cursor left 21 cells
> > 
> > The \E's are escape characters.  unmap renders all of the control 
> > characters with either ^'s or \'s (it also makes a new line on each
> > escape character, to make it readable).
> > 
> > Referring back to your trace, I notice that the trace-level should be
> > high enough, but I don't see PutAttrChar lines in that.  I'd expect
> > to see those immediately before individual characters are emitted.
> > 
> > Sometimes I add temporarily another call to _trace there to show
> > the actual bytes that are written - which is what I was considering
> > advising you to do to explore whether ncurses is emitting the ^@'s.
> > 
> > Right now, I'd suggest checking to see if PutAttrChar (in ncurses/tty)
> > is really being invoked, or if there's some bug in the tracing, etc.
> 
> Wow, that is strange. I get into PutAttrChar when I type normal
> characters. However, when I move the cursor, I don't get into
> PutAttrChar. I wonder why.....
> 
Wait, when I move the cursor with the mouse, I always get back the same
data from readline, and pass the same data to ncurses. The only
difference is, I move the cursor and call
        curs_set(1);
        wmove(scr->win, height - 1, scr->current.pos % width);
So, ncurses probably optimizes away the fact that the screen needs a 
redraw because it got the same data for the same line. 
What do you think?

Bob Rossi