Strange Problem

New Message Reply Date view Thread view Subject view Author view

Dave Russell (drussell++at++ist.ucf.edu)
Wed, 25 Feb 1998 14:08:39 -0500 (EST)

We have some code written in Performer 2.0 (can't remember the specific
revision) running on a couple of different Onyx RE's. The problem feels
like a memory munging thing since we haven't really changed the particular
part of the code where things are failing. The application starts by
loading two basic models. We've implemented a terrain following
intersection test to keep the eye height at a specific level of the
terrain, and this works find at first. At some point later in execution
other models are loaded with their associated controlling software
objects. If they are loaded within the first couple of frames, everything
seems to work fine; however, if we wait a little while to load the
additional models (several seconds), the next call to pfNodeIsectSegs
takes a VERY long time to return.

Even worse than that, the returned pfHit has a PFHIT_FLAG value of 127
which claims to have a valid intersection. Unfortunately, the returned
hitpoint has nan's for coordinates. the terrain following happily uses
this to adjust our viewpoint, and everything then really blows up. In
looking at the segset passed in to the pfNodeIsectSegs call everything
looks fine prior to the first call that takes several minutes to process.

It feels like the somewhere in the construction of our secondary models,
we're really messing up memory somewhere, but I would expect the segset to
be the culprit if we were stomping on some of our own memory. could we be
stomping on something that Performer has set up that would cause the
really extended intersection traversal?

I realize this one is not going to have a really easy answer, I've been
banging my head against the wall for nearly 15 hours on the problem and
really haven't made any progress. Any help would be appreciated if you
have suggestions. Otherwise, just ignore my rantings, and thanks for
letting me vent.

David Russell |
Visual Systems Lab | Static worlds breed
Institute for Simulation and Training | static minds.
|
drussell++at++ist.ucf.edu | CHANGE YOUR (virtual) WORLD!

=======================================================================
List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
            Submissions: info-performer++at++sgi.com
        Admin. requests: info-performer-request++at++sgi.com

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:56:53 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.