CLIM mail archive


Re: Moving objects and incremental redisplay

    Date: Tue, 28 Sep 1993 18:20 EDT
    From:     Nichael Cramer <ncramer@BBN.COM>

    Scott McKay <> writes:
	>I wrote:
	A heads-up:

	There seems to a problem in the CLIM if you move presentations around (this
	certainly true in ACL-CLIM, don't know about the others).  Specifically the
	standard, default output-history gets confused (this is particularly a
	problem when presentations overlap and/or are at the same vertical

	It seems to help replace the output history in your window with something
	like the following:

    >Yeah, the coordinate sorted set output history is pretty un-robust.  I
    >wonder if it should no longer be the default.  But I didn't really trust
    >my implementation of R-trees (it's *very* hairy), and did not dare to
    >make it the default.

    This is good to know.

    I just suggested R-TREES because someone else had suggested them to me when
    I ran into the above bug with coordinate-sorted-sets.

    Can you know suggest some type of output history that would be more robust?
    (In my particular case, it needn't be anything especially fancy as I can
    all but guarantee that none of the presentations overlap).

Actually, I don't know for a fact the R-trees are broken.  It's just
that they're hairy and hadn't had a lot of use, and it seemed safest to
err on the side of caution.

    (BTW, I may be having a similar problem with the R-TREES-OUTPUT-HISTORY.
    Occassionally when I remove large numbers of presentations one at a time it
    seems to run off the end and get a something like a "POP-POINTER = 0"
    error.  I've not reported this because I can't reliably reproduce it and as
    a consequence I haven't convinced myself it's not a bug in my code.)

Hmm, odd.  I suppose there could be a bug here in tree balancer that
gets called during deletion, but I dunno.  If you can come up with
anything reproducible, that would be great.

Main Index | Thread Index