gaaclarke

> Here you seem to have the same named argument twice. Because cond and !cond are (presumably) mutually exclusive, you'll only ever end up with one argument. Until cond changes...

I'm not sure if this case requires a test exemption since it has a test but it exists in the other repo.

For asset loading the time change for a 1MB payload is: 543.29 µs to 339.95 µs (37%) Benchmark: https://github.com/flutter/flutter/pull/105982

I had a chat with @jonahwilliams and @dnfield. We decided: 1) Having a mutable ByteBuffer that crashes when you write to it is unacceptable 1) Having a *private* channel that...

Hey, can't the operating system do copy-on-write with mmap? https://pythonspeed.com/articles/reduce-memory-array-copies/ This would be something to investigate.

> Hey, can't the operating system do copy-on-write with mmap? https://pythonspeed.com/articles/reduce-memory-array-copies/ This would be something to investigate. Looks like we are already doing copy-on-write (https://github.com/gaaclarke/engine/blob/a236ed5db78c665eb66de540a4fcda57c7862c71/fml/platform/posix/mapping_posix.cc#L75:L75), so this probably isn't a...

I was able to verify that copy-on-write works, and it's not what we are currently doing. I had to apply the following patch: ```diff diff --git a/assets/directory_asset_bundle.cc b/assets/directory_asset_bundle.cc index 7896756614..c0d0e0d844...

I filed a bug for the fuchsia failure: https://bugs.fuchsia.dev/p/fuchsia/issues/detail?id=102798 Since it works in darwin and android it seems to be a fuchsia bug. We can cut them out of this...

Okay, I'm putting this back into review. I've bubbled up the `GetMutableBytes` from `FileMapping` to `Mapping` and in the cases where we have access to the mutable bytes, I pass...

Sorry, pulling this down from review yet again. While it does functionally what I wanted, I've lost the performance benefit to platform channels on iOS when compared to previous versions....