Re: More enhancements to fringe bitmaps.

[Kim F. Storm]

YAMAMOTO Mitsuharu <address@hidden> writes:

> >>>>> On 18 Feb 2004 23:04:15 +0100, address@hidden (Kim F. Storm) said:
> 
> > I just installed a better fix.
> 
> It works fine.  Thank you very much.
> 
> So, the remaining problem is the one related to the window splitting.
> I believe the problem is specific to Carbon Emacs, but let me
> summarize how to reproduce it here.

I will think about how this can happen.  But it very much sounds like
a flush is missing somewhere (which may also be why the splash screen
does not work on Carbon).

In any case, while fixing the previous problem, I noticed that the
handling of the phys_cursor has some subtle problems, so I will try
to fix those first.

> 
>   (1) Launch Carbon Emacs (where blink-cursor-mode is disabled by
>       default).
>   (2) M-<, C-e, and hit the space bar four times to move the cursor to
>       the right fringe area of the first line.  (Now correctly
>       displayed as a filled cursor.)
>   (3) C-x 2  -> All the cursors disappear.
>   (4) C-x 1  -> The cursor comes back to the right fringe just after
>                 C-x is typed.
>   (5) C-x 2
>   (6) C-a    -> The cursor in the first window goes to the beginning
>                 of line, but the cursor at the fringe remains to be
>                 displayed.

emacs probably thinks it's already displayed...

>   [The following does not depend on the previous sequence.]
>   (7) C-x 3  -> The vertical line separating windows is not displayed
>                 until some event (e.g., keyboard input) occurs.

sounds like a flush is missing.

> 
> If blink-cursor-mode is enabled, (3) and (7) are solved after the
> first blink, and (6) does not occur unless you type C-x 2 C-a quickly
> before blinking starts.

because blink is followed by a flush?

> 
> Although I'm not sure it is related to the above problem, I noticed
> that x_make_frame_visible in Carbon Emacs is different from the ones
> in other platforms.  It does not wait for a window system event that
> tells the frame becomes actually visible to come.  I guess this is the
> reason why the "fancy splash screen" is not displayed in Carbon Emacs
> with the image support patches.

or output is not flushed correctly?

maybe you can compare the x and mac ports to see if there are
any differences when/how things get flushed...

Also, XFlush and x_flush in macterm.c are defined differently
on Carbon.  Maybe one of them has a bug?


> 
> As for the vertical line separating windows, I noticed that it makes a
> flicker in the X11 version if blink-cursor-mode is enabled.

I have seen the vertical line flicker on X myself; I'll look into
that as well.

-- 
Kim F. Storm <address@hidden> http://www.cua.dk