Angus Dorbie (dorbie++at++sgi.com)
Fri, 29 Jan 1999 14:51:12 -0800
In effect you were porting a single threaded application which had been
written over a period of 3 years to a multiprocessor system.
The realloc(0) problem was probably related to making draw calls in the
application.
It's fair to compare your porting effort to the difficulty you'd have
had efficiently exploiting your OCTANE multi-processing without
performer. Despite your experience I expect that Performer was in fact a
big help to you in this respect.
I think your suggestions are great, I expect everyone who has
implemented a sizeable performer application has encountered similar
problems and solved them independently.
Cheers,Angus.
Marc Erich Latoschik wrote:
>
> Hi pfFriends,
> Some of you might have read the postings i made in the past about having
> major trouble porting a performer application from irix6.2 on Indigo2's
> to my brand new octane running 6.4.
>
> And you might have similar problems i had, so here is the current state:
>
> Fortunatly i can tell you that all my problems are solved by now ;).
> Unfortunatly, for that, i have made a complete rewriting of the code!
> And i'm still stuck with guesses what could have happened (read the old
> mails),
> that in fact is one of the things i hate, cause there is still this
> little
> uncertainty in the back corner of my head, you might know that ;)
> feeling.
> Anyway, it runs ok,
>
> but during the whole process some questions about Performers C++ API
> arose
> too, in first tries to use Performer, most people will use a copy and
> paste
> solution from examples in pguide etc (most C API). That is how my app
> started back 3 years from
> now. But after a while you might experience more and more that the
> complexity of your programs will not work in any case with that simple
> solution.
> The OO paradigme seems a good way to come around it. And Performer
> supports
> it with it generic C++ API.
>
> But what is painfull is that there is only one chapter in the pguide
> about
> how to subclass pf-nodes, what is in fact the first thing that comes to
> ones mind. But to really unleash the object-oriented power you really
> want
> more, well i did at least. And to get more, basically you can try and
> error
> subclassing anything else, most of the time not beeing shure if this was
> the easiest
> way to solve a given problem.
>
> Maybe here are more people that want to share their ideas and solutions
> on C++ program design on a problem based basis? To name just a few i was
> dealing with:
> - How to implement specific object features, animated nodes, kinds of
> animation and more
> - How to implement a generic view class that supports all kinds of
> stereo.
> - How to have abstract classes that encapsulate different projection
> medias
> (HMD,s big screen's, and more)
> - How to achieve online switching between different view models
> - ...you name it...
>
> If this will be done in an open discussion, i think lots of people will
> think: " oops, thats a good idea, i should have done that"
> in contrast to try and error ;)
>
> what do you think?
>
> --
>
> bye
> Marc
>
> **************************************************************************
> Marc Erich Latoschik, AG-WBS, Technische Fakultaet, Universitaet
> Bielefeld
> Universitaetsstrasse 25, PO 100 131, 33501 Bielefeld Room
> M4-122
> Tel: (0521) 106 2919, Fax: (0521) 106 2962,
> marcl++at++techfak.uni-bielefeld.de
> http://www.TechFak.Uni-Bielefeld.DE/techfak/persons/marcl/
> =======================================================================
> List Archives, FAQ, FTP: http://www.sgi.com/software/performer/
> Submissions: info-performer++at++sgi.com
> Admin. requests: info-performer-request++at++sgi.com
-- "Only the mediocre are always at their best." -- Jean GiraudouxFor advanced 3D graphics Performer + OpenGL based examples and tutors: http://www.dorbie.com/
This archive was generated by hypermail 2.0b2 on Fri Jan 29 1999 - 14:51:20 PST