Hi Martin et. al, On the topic of masks and the debate that Tom pointed to. At the time said I would let the issue lie and return to it once I'm come to conclusion once things had settled. It's one of these issues where it's six of one and half a dozen of the other. Time has gone on and your post has spurred me to finally make my mind up.
Masks to be unsigned int or int? On considering all the different possibilities and issues I've come to conclusion that types for masks where enum's are the most natural way of specify the options then int's should be used. Traditionally the OSG code bases uses unsigned int in this situation, and I've always felt it more natural that a mask would be an unsigned int, but... enum's are int's and with compilers getting more a more picky about types the use of unsigned int for masks that we use in conjunction with enum. Either we'd have to ditch the use of enum's in this context and adopted #define or static const unsigned int usage in it's place, or adopt int's. The later seems like the line of least resistance, and easist to maintain long term (as it's the most minimal code wise) and the easiest for OSG users as it should avoid the need to introducing casting that users previous had to do to avoid some warnings. I have now gone through the OSG headers and source files and updated them to using int's for the mask types where appropriate. One place - osgSim::ObjectRecordData I've kept unsigned int as this matched the OpenFlight file format that it was mapping, and have replaced the previous enum with a series of static const unsigned int declarations in the headers. This combination avoids warnings. These changes are now checked into svn/trunk. Fingers crossed there won't be any build breakages, and hopefully should address some of the warnings - it's fixed some for me ;-) Cheers, Robert.
_______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
