Re: moving point over invisible and intangible text

[Richard Stallman]

             I believe that Stefan's implementation
           and the invisible-intangible feature are pretty similar except for
           their treatment of stickiness as to which side point is going to 
move.

       What is the difference in behavior, precisely?

    I originally misunderstood the difference in behavior, mistakenly
    thinking that it had something to do with stickiness.  Apparently, the
    only difference is that Stefan's implementation only adjusts point
    after an interactive command, whereas the alternative parallel
    implementation adjusts it inside non-interactive code as well.

Are they exactly identical in terms of which positions of point
are not allowed, and which positions they move point to?
If not, what are the precise differences?