Re: Unexpected hiding of identical images

[Richard M. Stallman]

    What is the problem of one vs. multiple chars?

The issue is that there's no possibility of killing or yanking part of
one character--so that case is easier.  The case of
auto-image-file-mode is harder.

Isn't auto-image-file-mode more or less obsolete now?  I wrote
image-mode to replace it.  But image-mode probably has the same
difficulty, as regards this issue.

    Either it is put on a single character, e.g. by insert-image, or on
    the full string representing the image data, e.g. by auto-image-file-mode.

For full correctness, it needs to treat partial and complete yanks
differently, perhaps somehow to recognize the complete text.

Here's an idea.  Put one special property on the first character of
the image, and another on the last character.  These should enable
the yank handler to DTRT in all cases.

What do you think?