Let's see if I can answer some of your questions and clarify some statements. First of all, "annual glare" can mean a whole bunch of things. What's done in OpenStudio using Radiance as the simulation engine is a very crude riff on Daylight Glare Probability called "DGP-simplified". This is based on illuminance, and as such, is using an integrated value for incident flux at the observer's eye -- it doesn't take into account the size or position of any sources. It's also only been shown to give good agreement with research test subjects when direct sun is always blocked or highly diffused to the viewer, which up 'till only recent versions of OpenStudio (~1.8.0) made it pretty difficult to rely on this DGPs for much. This has changed somewhat in the new OpenStudio 1.9.0.
In OS 1.9.0, you can actually simulate (with Radiance) the effects of venetian blinds, shadecloth (e.g. Mechoshade products), and even daylight redirecting louvers (e.g. Lightlouver), making the DGPs values a little bit more meaningful, albeit with the results being derived from low resolution BSDFs. The bigger news with respect to this BSDF business is that now in 1.9.0 users can assign different shading controls to the model and the effects of having different controls for each facade orientation are calculated and blended so that the annual daylight results from Radiance are spatially coherent.
In keeping with the "there's no such thing as a free lunch" principle, the BSDFs used in OpenStudio are generic, were derived using the low-resolution Klems sampling basis (i.e. the hemisphere is subdivided into 145 equal patches), and are included in the OpenStudio application; they are not on the BCL. The user is limited to these included BSDFs in 1.9.0, you cannot assign your own at this time. That said, we think this is a big improvement to what you can do with OpenStudio and Radiance for this release, and we will be pushing documentation on these features live in the next couple of days.
