No modern GPU hw support double vertices, so we end up translating them to floats internally in the driver. There is probably little reason to use GL_DOUBLEs. http://cgit.freedesktop.org/libreoffice/impress/tree/slideshow/source/engine/OGLTrans/OGLTrans_TransitionImpl.cxx
B2DTuple is using doubles, so I'm guessing if we want to present floats to the opengl layer we'd probably have to do something like turn B2DTuple into a template and use B2DTuple<float> in impress/slideshow/source/engine/OGLTrans to avoid ending up just doing the same double->float conversion ourselves
I think this is in Rodo's area? Reassign back to the list if not.
Back to list due to facts Very old Bug report, not easy to understand for me what consequences "it" has.
Does this belong in UNCONFIRMED or should it be in NEEDINFO?
I'll set this "confirmed". This is a user-invisible, mostly plumbing-only kind of bug report. I doubt this really impacts end users directly.
Potential Easy Hack then?
Changed priority to lowest enhancement as Kohei has stated that the real impact will be minimal if anything at all
Ah well, so our B2DTuple are doubles and it would be pretty dramatic to change that to floats at this stage to use floats throughout LibreOffice and presumably converting them to floats just at the point of insertion into opengl is effectively the same thing that the driver does already and would gain nothing.