Po Lu <luangruo@yahoo.com> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>
>> That's the technical description of the implementation.  It doesn't
>> explain when  canvases can be useful and for what purposes.
>
> Canvases are useful for when I want to be able to control a portion of a
> screen dynamically, in a fast way, from Lisp code.  For instance,
> displaying a constantly changing bar chart or graph inside Emacs.
>
>> But the result of this painting is some graphical object, similar to
>> an image, that will be displayed within a buffer, right?
>
> Correct.
>
>> Did I miss the code that tells Emacs the click was on a canvas?  E.g.,
>> if you click on a canvas, what does posn-object return when passed the
>> click event as its argument?
>
> Oops.  I missed that.  Thanks for bringing it up.

Here is a version of the patch with several of the problems you
mentioned rectified.