Remove NodeId usage from ChromiumSenderThread
#356
Conversation
noituri
force-pushed
the
@noituri/chromium-thread-node-id
branch
from
af669d9 to
50d2589
Compare
|
|
||
| pub fn resize(&mut self, size: usize) -> Result<(), SharedMemoryError> { | ||
| // Releasing the current `Shmem` instance to ensure it does not erase the shared memory descriptor from the file system | ||
| // This is critical to ensure when a new `Shmem` is created at the same location, it doesn't conflict with the old descriptor | ||
| drop(self.inner.take()); |
There was a problem hiding this comment.
| drop(self.inner.take()); | |
| self.inner.take(); |
wouldn't this be the same?
There was a problem hiding this comment.
This whole line is not needed unless I forgot about something.
| } | ||
| } | ||
|
|
||
| // Create additional shared memory |
There was a problem hiding this comment.
what if the number of inputs is smaller, there is no cleanup of already allocated memory
There was a problem hiding this comment.
There is no cleanup since it requires synchronization. It will be added eventually
| frame.send_process_message(cef::ProcessId::Renderer, process_message)?; | ||
| // ----- | ||
|
|
||
| shmem.resize(size)?; |
There was a problem hiding this comment.
what if you only resize if shm is too small? Would this avoid potential crashes even without synchronization?
|
I need those changes for the other PR I'm working on and it seems everyone already reviewed it, so I'll merge it. Please take a look at the comments and if there is anything actionable there address that in a follow-up PR. |
Initially, when I implemented the web renderer I thought that it would be possible to have one web renderer that is used by multiple web views. In this PR I removed some of the unnecessary stuff that was added because of the assumptions I made when I was implementing it.
Issue: #352