Hi J-S, Your suggestion of a data file testing suite is a one I've contemplated for a while, and it's been raised in a bit of an ad-hoc way on osg-users over the years. Perhaps it's time for use to tackle this a bit more formally, wire it up with the CTest suite and make it a proper unit test suite.
Robert. On Thu, Feb 19, 2009 at 7:35 PM, Jean-Sébastien Guay <[email protected]> wrote: > Hi Robert, > >> Frustratingly now one seems to have tried out hdr imagery >> in those two months either...despite all the calls for testing. > > I can take part of the blame for that. I tried to show my Masters project to > some co-workers a few weeks ago, and compiled it against trunk (before the > push for 2.8). I saw that my *.hdr images (cubemaps, etc.) didn't load, but > I didn't have any time to chase it up and so just let it drop thinking it > had to be something on my end (on the assumption someone else using OSG > would have posted here if it had been a real problem with OSG itself). > > When the calls for testing 2.8 started, I should at the very least posted a > quick note that my HDR images didn't load but I didn't have the time to look > for the real cause, and suggested someone else at least try loading one HDR > image. Sorry, lesson learned. > > At the risk of starting another one of these subjects which don't get any > concrete action... Couldn't we just put together an automated script that > would try to load all the data formats that standard OSG with the bare > minimum plugins is supposed to load? For example, something that would use > osgconv to convert any readable image / model type to something OSG can > write (I guess tiff for images and .osg for models for example). The script > would thus not require any intervention, but examining / parsing the output > (perhaps with OSG_NOTIFY_LEVEL=DEBUG) would reveal these kinds of bugs at > the very least. > > If we were really thorough, the script could even test every possible > combination of reader - writer to make sure they at least read the data and > write it out. Even without checking the output, the notify messages would at > least flag any really serious errors like the one we just saw with the HDR > loader. > > We could run it right before a release as a minimal test. Apparently, we > can't assume even the large number of people using OSG will be using every > possible data format it supports, and automatic testing like this could give > us a bit more assurance that the basics work. > > J-S > -- > ______________________________________________________ > Jean-Sebastien Guay [email protected] > http://www.cm-labs.com/ > http://whitestar02.webhop.org/ > _______________________________________________ > 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
