You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Jun 18, 2021. It is now read-only.
We know that gfx-hal (with Rendy) provides a different set of trade-offs versus wgpu-rs. It makes the choice for library vendors to limit the future use of the libraries. If a library is simple enough that it could be using gfx-hal directly, and the author considers going this path, it means the wgpu-rs applications would not be able to use it, and visa versa. The issue is inspired by servo/pathfinder#213
It would be interesting to try to define a foreign library interface (FLI?), such that the user can unsafely provide the necessary glue bits in order for wgpu-rs to be able to use the logic bits of the foreign library. This could be enclosed into a "glue" library, e.g. "pathfinder_wgpu", which anyone could then use as a regular library when their graphcis stack is based on wgpu-rs.
The difficulty here is defining where the boundary should be drawn, and how to provide all the tracking info to wgpu. It might be infeasible completely, just needs to be investigated.
The text was updated successfully, but these errors were encountered:
We know that gfx-hal (with Rendy) provides a different set of trade-offs versus wgpu-rs. It makes the choice for library vendors to limit the future use of the libraries. If a library is simple enough that it could be using gfx-hal directly, and the author considers going this path, it means the wgpu-rs applications would not be able to use it, and visa versa. The issue is inspired by servo/pathfinder#213
It would be interesting to try to define a foreign library interface (FLI?), such that the user can unsafely provide the necessary glue bits in order for wgpu-rs to be able to use the logic bits of the foreign library. This could be enclosed into a "glue" library, e.g. "pathfinder_wgpu", which anyone could then use as a regular library when their graphcis stack is based on wgpu-rs.
The difficulty here is defining where the boundary should be drawn, and how to provide all the tracking info to wgpu. It might be infeasible completely, just needs to be investigated.
The text was updated successfully, but these errors were encountered: