Feature Request: Layout arguments as aesthetics

[billdenney]

I often use manual layouts. It seems like it would often be simpler to pass the layout arguments to ggraph() via an aes() rather than ... arguments to ggraph(). I realize that these aesthetics are categorically different than those used for the geom_*()s, so it could be confusing to the user, but in other ways, it seems like a logical fit.

Mostly, it is a feature for consideration that I think could help, but I would completely understand if it were not considered a good fit.

[schochastics]

As the developer of the graphlayouts package I would vote against this. While there are some parameters that are similar across different layout algorithms but overall, they differ significantly. As such, I think the ... is the most feasible approach, given also that most parameters have very good default values that do not need to be changed much.

[thomasp85]

Agree with David. Further, there is no established or obvious conventions over whether node or edge data should be referred when creating a layout

[billdenney]

Thanks for considering it. I understand the desire not to include it.