Weird behaviour

New Message Reply Date view Thread view Subject view Author view

Torsten Moeller (tmoeller++at++cgrg.ohio-state.edu)
Tue, 14 Mar 1995 01:59:17 -0500 (EST)


Dear Performer community,

I've been messing around with the following problem(s) for a few
days now and don't seem to make any progress. Although it might look
kind of trivial it is not (at least not for me).
To build my application I basically use perfly as a guide.
I am building the following hierarchy:

scene -> sceneDCS(for Trackball) -> A: dataDCS -> B: sceneGroup

sceneGroup are a bunch of little objects which construct a
building. Now there are several weird things happening:

I included A to correct the building geometry I am working with (it's
tilted).
a) Since I couldn't use pfFlatten to increase performance (see previous
        post) I am using a SCS/DCS and did some performance tests.
        I found out, that a DCS is a little faster in performance.
        So why would anybody ever use an SCS at all ?
        Also there must be an error in the man page for pfSCS as well
        as for pfDCS. To increase performance I of course set the
        bounding box to PFN_BMODE_STATIC. To do that, the man page for
        pfSCS suggests pfNodeBSphere(scs, PFN_BMODE_STATIC), but here
        is the bsph argument in the middle missing. The man page for
        pfDCS suggest (supported by the explanations in the man page
        for pfNodeBSphere) pfNodeBSphere(scs, NULL, PFN_BMODE_STATIC).
        But when I try that my data structure simply disappears.
        I finally got it to work doing the following:

        pfGetNodeBSphere(dcs, &bsphere);
        pfNodeBSphere(dcs, &bsphere, PFN_BMODE_STATIC);

b) I also tried to change the order of A and B. To my surprise
        the output on the screen was tilted. To get the same effect
        on the screen I had to change coordinates from x to y and
        from y to z and from z to x. Do I miss here something? Does
        anybody know why this happens?

c) If I get my program to run, I can run it on a Crimson an Indigo
        and an Personal Iris, but NOT on an Onyx. After reading the
        database it just simply quits with a message:
        "Performer Notice (9): Caught SIGCHLD. Exiting due to death of
        child with pid xxxx"
        Now I got former versions of my program to run on the Onyx,
        as well as perfly and other demo programs.
        Did anybody experienced similar behavior?

d) Sometimes the Gui will be totally wrong. First of all the entries for
        x,y,z and h,p,r are totally wrong. Although lighting is set
        to 'ON' it starts with 'OFF'. I will get buttons for Anti-
        aliasing but the Crimson I am using does not have a Reality
        Engine, and therefore shouldn't show this option etc.
        I usually get rid of this problem, including a printf command
        somewhere in the code and recompile or something like that.
        It does sound like a badly used pointer error, but I looked
        over my code a zillion times for such errors, but just can't
        find anything.
        Can anybody explain why this happens?

Well, I hope somebody is able to give me hints what I could check or
where I could look for errors in my program. At this point I just
get a little frustrated, because I don't know anymore where I could
look for an explanation (and possible) fixture of the 'phenomenas'
presented above. I hope I don't bother anybody with these questions,
but I am the only one in our lab dealing with Performer, so there is
nobody else to share ideas with.

Any help is greatly appreciated,

                                        Torsten Moeller.

-- 
Research Assistant   			ACCAD
tmoeller++at++cgrg.ohio-state.edu		Ohio State University

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:51:04 PDT

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