Hi Mathias, I have gone ahead an merged your changes. I'm actually tempted to go further and remove the whole dual repeat/clamping texture code so that there is just one texture as things stand now the clamped texture is totally redundant and potential confusing for future maintenance.
There other thing I wonder is whether turning clamping on is something might something you can autodetect by looking at the texture coordinates on the geometry to see if it stays in the 0.0 to 1.0, if so then clamping would be appropriate. Thanks, Robert. On Sun, Oct 26, 2008 at 12:04 PM, Mathias Fröhlich <[EMAIL PROTECTED]> wrote: > > Hi Robert, > > I am harder to reach this week. Sorry ... > > On Freitag, 24. Oktober 2008, Robert Osfield wrote: >> I've just reviewed your changes, but hacking stuff out and disabling >> the ability to not do clamping is just going fix some models for some >> users and break some models for some users. It might be that more >> users will be assisted by this change than hindered, but still the >> original clamping change was not put in for no reason so we do no at >> least some users will have problems once I apply this change. Then >> perhaps a few more months down the line we'll get another submission >> that flips it back the other way again and will just keep flip >> flopping between them. >> >> I think the right thing to do is strike up another discussion on >> osg-users with a call for models that demonstrate the different ends >> of the spectrum on this issue, and then work to get changes to the >> ac3d plugin then ideally load all of them as the the various end users >> expect them to without any additional input form them. If it is a >> case that some .ac models don't follow the spec then we could just >> declare them erroneous, or add an optional parsing support for the non >> standard interpretation. > > I believe I have already done so. We have previously had a thread on osg-users > regarding that. I have put John who has requested that original offending > change directly on the cc list. I got no answer at all from him. > He requested a change which turned out not to match ac3d's behaviour nor their > specification for the ac files. This change came in only a few weeks ago. It > was a patch from him that I reimplmented in a way that shared much more state > attributes than his patch did. > IMO I must appologise for supporting his change request without looking at > ac3d's behaviour and specification. > > So this current change just reverts to what is specified, matches ac3d and > restores a behaviour that was ok for many people for at least two years. > > If John wants to have a diferent texture repeat, he can install a post read > callback or something like that. I have already asked what he likes to have > or if we want to have an option for that loader but we did not get any > response. > In this case I think we can revert to the specified and ac3d matching > behaviour. > > Greetings > > Mathias > > -- > Mathias Fröhlich, email: [EMAIL PROTECTED] > _______________________________________________ > osg-submissions mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org > _______________________________________________ osg-submissions mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
