I work for Autodesk's AutoCAD division. I don't actually work on Lightscape codebase but I do know a little of how they do lighting and radiosity. Here are my two cents:

Generally speaking, most 3D programs more or less calculate specular lighting this way:

S = Sm * Sl * Pow (R,Sh)

where S is the final specular intensity, Sm is the material's specular value, Sl is the light source's specular, Sh is the material's shininess, R is the reflection factor based on the geometric positions of the lighting source and the face orietation, and Pow() is the power function.

So, Lightscape is doing the right thing. It's only natural that S would be 0 if the material's specular value is black (i.e. 0), regardless of the shininess.

In other words, the word "反射率" that you used in your message is confusing in the sense that it could mean either the material's specular value or its shininess, with wildly different meanings. (I may be wrong here, for I rarely read technical writings in Chinese and I don't really know how they translate these terminlogies.)