🚀 go-pugleaf

Hi, I've been bothered by something for a while and I wondered if anyone could
give me a legit explanation of why exactly it's necessary.  First I'll describe
the symptoms.  I have, say, two windows on the workbench screen, both the size
of the screen.  I open up a very small window (like the virusx title-only
window) and start dragging it around the screen, dropping it off more or less
randomly, and then moving it again.  After about ten or twenty times, the
delay between when I let go of the button and the time I have control of the
mouse again is very high, more than a second.  Now, if I try to flip back and
forth between the other two windows, it takes a long time and it's easy to see
how the system has divided the whole thing up into rectangles.
    Okay, now it's pretty obvious what's going on here, so I won't embarrass
myself by misdescribing it, but my question is the following: why?  Is it too
costly to try to consolidate some of these areas after every move?  If this is
true, couldn't it be set up like lisp garbage collection so that it only went
through the whole process, say, every so many rectangles or perhaps according
to system performance?  Or am I misunderstanding something that makes it
clearly impossible to do what I want in any sensible way? As it stands, it's
somewhat annoying for people like me who have the nervous habit of idly
dragging small windows around the screen while they're thinking.  Thanks for
any info,

                                                   -Dan