Re: lynx-dev [BUGS 2.8.3.dev4] partial, off-by-one

[Klaus Weide]

On Sun, 18 Jul 1999, Ilya Zakharevich wrote:

> b) In a lot of situations I see the "-more-" prompt written one char
>    too early: first dash is on the last column of the last-but-one
>    line.  The resulting scrolling messes up the display (AHA, probably
>    the opposite: it is written on the last position of the last line,
>    but the scrolling moves it to the previous line).
> 
>    The simplest way to reproduce it here is to do
> 
>    h ^L
> 
>    But I also see it after middle-mouse-click===then
>    right-mouse-click-outside (to remove the menu), after printing, and
>    after downloading.
> 
>    Local config: EMX port in a VIO window, gzipped help, latest ncurses.
>    Reproduces with TERM=ansi as well.

If your problem is that your curses library cannot deal with
positioning on the last screen position without somehow leading to
scrolling, AND you cannot avoid that with the right setting of
-show_cursor, you should probably talk to "HV"; in MainLoop.c:

    /* turn off cursor since now it's probably on statusline -HV */
    move((LYlines - 1), (LYcols - 1));

I remember having had a discussion about this, it should be protected
by 'if (!LYShowCursor)' and/or '#if defined(FANCY_CURSES) || defined(USE_SLANG)'
as is done in some other places.  I don't remember who HV was, VH was that
you?

> c) Looking through src/LYStyle.c:parse_style I observe conversion to 
> lowercase:
> 
>    first it is done manually char-per-char, then an additional call to
>    strtolower() is performed, but to be extra sure, the results are
>    compared via strncasecomp().

That's to make *really* sure those letters are really as small as
possible. :)

>    BTW, why 'n' in strncasecomp()?  Is not it leading to wrong semantic?

Yes, it seems so.  That's all more or less inherited from Rob's original
code.  Maybe at some point it wasn't clear that 'element' was only
the "element" and nothing more.

  Klaus