Join GitHub today
GitHub is home to over 50 million developers working together to host and review code, manage projects, and build software together.
Sign upRender crates #236
Render crates #236
Conversation
|
Rather than adding a glutin dependency to the media crates, I think we should define a trait that provides the necessary APIs. This can then be implemented by the servo embedder, which will allow us to more easily integrate with platforms where we don't use glutin (like Android and magic leap). |
ceyusa
force-pushed the
ceyusa:render-crates
branch
3 times, most recently
from
7141f15
to
0f144f5
|
Just tested the branch in Mac and Win10, and in the first works, in the last compiles, but I got a strange webrender error. Removing WIP. |
|
@ferjm: r? |
|
\o/ Great work! I know the current state of documentation for this crate is not very good, but I think this is complex enough to deserve a little bit of documentation ;) Thanks for fixing this! |
This object is in charge of two tasks: - Create the video sink - Create the frames to push to the Renders The purpose of this design is to handle GL by another create (in order to workaround servo/servo#22944 (comment)), and that other object will proxy these tasks if available in the target OS. This patch removes temporarly the GL support in player.
|
@ferjm: r? |
|
@bors-servo r+ |
|
|
Render crates This PR creates two new crates inside gstreamer backend: - A render crate which only defines a trait for different renders for different OS - A render object helper that will setup the rendering bits for the player - A unix-render crate which implementes the render trait for Unix-based GL systems This will fix the problem with cargo with features per target. It is not well tested yet. And the GL implementation still has some missing bits. I would like to propose a new create depending on glutin to extract the GL context and display for different platforms used by the possible gl renders.
|
|
ceyusa commentedApr 6, 2019
This PR creates two new crates inside gstreamer backend:
This will fix the problem with cargo with features per target.
It is not well tested yet. And the GL implementation still has some missing bits.
I would like to propose a new create depending on glutin to extract the GL context and display for different platforms used by the possible gl renders.