Re: undo weirdness with insert-file-contents

[martin rudalics]

>>-       current_buffer->undo_list = Qt;
>
> But this patch is clearly problematic, if one believe the comment: it
> would cause the generation of undo info during the decoding step.
> I understand that this info will be thrown away later, but we don't want
> to pay the corresponding price.

The comment is mine, don't trust it.  Emacs 22 records undo information
all through the decoding step.  That's what I wanted to avoid.  But I'm
afraid two insertions get collapsed into one by the undo recording
mechanism and I'm comparing the wrong starting positions.  I don't yet
understand how this happens.