[Bf-committers] Blender developer meeting notes - 2019.10.14

[Gustavo Niemeyer]

Hi Blender devs,

Sorry if this is not the right place to discuss this:

On Tue, Oct 15, 2019 at 1:14 AM Dalai Felinto <dfelinto at gmail.com> wrote:

> Announcements
> =============
> * Jacques Lucke presents a new design for particle nodes [1].
>

Is there something I might read about the rationale motivating unrelated
data "types" to be unified in a single socket, and the more "procedural"
take on the nodes rework?

This screenshot is what made me wonder about the design:

    https://wiki.blender.org/w/images/e/e9/Particle-nodes-05.jpg

>From past experiences on such systems, we usually see a slightly more
declarative fashion. For example, "Normal" from the surface would plug
directly into a "Velocity" socket on "Mesh" or something similar. It sounds
like this new design is going for power and flexibility.. the nodes are
representing code blocks instead of information flow. The price of that
will be additional complexity and making the node graph harder to read, I
suspect.

Anyway.. I appreciate such node systems in general, so I'd love to read
something about the background for such choices, if it exists.


gustavo @ http://niemeyer.net