Hi Stephen, Glad to hear that you've fixed the crash and improved performance. As to what the actual problem is... well I don't have grounds to suggest anything in particular, if the problem is mysterious at your end its 10x more so at this end - I haven't see your app let alone had a chance to debug it.
With this type of problem I'd normal point to likely areas - OpenGL driver bugs, memory management bugs, or uninitialized memory bugs. Since you haven't specifically fixed the bug, one can't point to one of the three above or to anything else that might have been amiss, which is shame as one can't learn much from the experience save for "weird stuff happens at sea" ;-) Robert. On 12/15/06, Stephen Northcott <[EMAIL PROTECTED]> wrote:
Hi all, Firstly, I haven't taken the opportunity to say this before.. But feel it is overdue.. A big big thank you to Robert (particularly) and everyone involved in OSG. It has become an invluable tool for so many projects for me, and the 'familly' that surrounds OSG seem to always be available to help. This place is an exceptional example of dedication and cooperation in the Open Source community. And I personally have re-learned and learned many things using OSG in the last 18 months or so. Fantastic! I digress.. :-) I posted a query to Robert a while back enquiring if he had any idea why a piece of code that dynamically generated a model from procedural sets of vertices, primitive sets, and normals and so on would cause an indexing error and crash on the PC and not on the Mac.. Well I found a solution (working on something else in the same project - as is often the way), and wondered if Robert, or anyone had any comments, or if this is perhaps a bug in OSG. I found that when I was creating a large object dynamically from vertices and primitive sets that I was in fact generating one primitive set per QUAD and adding all those primitive sets to the single geometry object... Not very clever, and also not very fast! So I was glad to find the mistake and correct it.. Boy did things speed up for me when I did that! So I altered this code so that it was just one primitive set encapsulating the entire set of vertices. This piece of code is also at the very heart of the code that causes the crash on the PC, so I then went to see (in typical programmer fashion) if the problem had magically gone away.. The crash on the PC has indeed gone away... and I can get it straight back again by changing the code back. My questions are.. 1. Is what I did originally illegal? Whilst it's not very bright I can't see that it is illegal. I can see some examples, pretty esoteric I agree, but nonetheless some examples of where you may want to grounp many small sets of QUADS in individual primitive sets. So should this cause problems? 2. Is it perhaps simply a memory problem. i.e. Is what I was doing very costly on memory and that causes the crash when it overflows.. 3. Can anyone else replicate this crash? i.e. About 6 - 8 15,000 polygon models, all made of quads, with normal and colour attribute arrays, all with parent primitive sets for each QUAD. Thanks to anyone who wants to comment on this.. Kind regards, Stephen. _______________________________________________ osg-users mailing list [email protected] http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
_______________________________________________ osg-users mailing list [email protected] http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
