Rafal Jaroszkiewicz

Allowing the derived class B to access internal nodes of A creates tight coupling: if the internal network of A changes sufficiently, it will break B. Ie, In my view,...

For sure, I can see the appeal of inheritance producing minimal XML code for the derived shaders (like name="ND_UsdUVTexture" and its name="IMP_UsdUVTexture_22"). It's just that it can come at the...

For the future version of MaterialX, we were considering to allow connecting a uniform string inputs (eg, of an image texture node, equivalent to USD UV Texture) to a **uniform**...

No major objections. Sounds like a reasonable approach.

Looks really good to me! Just some thoughts on the connection: ``` [output="output on childNG"]> ``` It's logical to use `nodegraph=` to specify the `"childNG"` as input, since node graphs...

Looks good! Great improvement over `ramplr` and `ramptb` shaders. It supports a constant and linear interpolation, and I wonder what it would take to also support a smooth interpolation (eg,...

It seems to me that the build-time solution is a stepping stone towards the run-time solution. Ie, the work on real-time will require code that bakes out the concrete `nodedefs`...

Anders' [PR](https://github.com/ld-kerley/MaterialX/pull/1) looks really good! Thank you! Seems like the spec proposal could be based on that.

One thing that caught my eye is how the `uniform="true"` attribute of an input element forces a separate template. It would be nice to have just one template, but I...

> An extension of the existing XML syntax, that is processed and then removed during the build stage. That seems to me like a good solution. It shields the integrations...