On 5/11/2011 1:46 PM, Ryan Pavlik wrote:
On Wed, May 11, 2011 at 1:27 PM, Paul Martz <pma...@skew-matrix.com <mailto:pma...@skew-matrix.com>> wrote: On 5/11/2011 11:20 AM, Ryan Pavlik wrote: On Tue, May 10, 2011 at 8:24 PM, Paul Martz <pma...@skew-matrix.com <mailto:pma...@skew-matrix.com> <mailto:pma...@skew-matrix.com <mailto:pma...@skew-matrix.com>>> wrote: On 5/10/2011 7:11 PM, Paul Martz wrote: On 5/3/2011 3:02 PM, Ryan Pavlik wrote: I'd like to propose my fixes that I sent in on -submissions pre-2.8.4: all except the VC2010 build fix still apply. They are these two: Improvements to osgconv: https://github.com/rpavlik/osg/compare/OpenSceneGraph-2.8...improve-osgconv (I've also forward-ported these to trunk: https://github.com/rpavlik/osg/compare/master...improve-osgconv-trunk ) So, if I understand correctly, you did not post this change to osg-submissions for inclusion in OSG's svn trunk, is that correct? Or, you did post them, but they are awaiting action by Robert? Has anyone other than you tested the changes? What, exactly, are these improvements to osgconv? (No, I haven't looked at the code -- yet.) Having reviewed the code, I see two changes: 1) A simple change to an error message in osgconv.cpp. Looks straightforward. 2) A change to the image file name format as output by the dot OSG plugin. Not sure I understand this one, but it doesn't appear to be an "improvement to osgconv". Did you include this by mistake? -Paul No, the .osg plugin modification essentially only affects use of osgconv. What it does is, if the OutputTextureFiles option is turned on (such as by passing -O OutputTextureFiles to osgconv), ...or anytime "OutputTextureFiles" is present in the Options string. This functionality is orthogonal to osgconv. Any application can access it anytime they write a dot OSG file. This change will affect more than osgconv. -Paul Well, yes. In retrospect, I didn't name the patch very accurately given its contents - I created it while working on osgconv which is presumably why I mentioned osgconv in the name. In any case, we can consider it as a separate patch, to the .osg writer plugin. My intent was to make saving models with textures to OSG format work better. In all of the following cases, insert the qualification "with OutputTextureFiles enabled" - behavior does not change unless this option is added. Before and after the patch, loading an .osg model in the place where it was initially saved works. After the patch, however, the textures go to a subdirectory named in a way that clearly links it with the model that uses it, and loading the saved model from another location as long as the textures were also copied/moved works correctly. The only case I can see that worked before and won't work after, is if the .osg file alone is moved/copied somewhere else, but the textures remain in the originally-exported location. (This is due to the fact that it now exports relative paths.) As a useful side-effect (for privacy/security) of using relative paths, .osg files saved with this option turned on don't reveal details about the file organization of the computer they were saved on. Would this be an improvement that would better be separated into a distinct option string like OutputRelativeTextures or something like that?
Yes, that's the way I'm leaning now. Your patch changes behavior, and would break any apps or utilities that depend on the current behavior. I'd be a lot more comfortable with this if it were a feature addition rather than a change to existing functionality.
Do you think "OutputRelativeTextures" is a good name for it? If so, I can rewrite the submission locally (but you'll need to test it).
-Paul _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
