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
Just want to play devils advocate here: Is this truly going to be worth moving away from tokio for the runtime? Tokio seems to be the de-facto runtime.
The OpenDAL analogy is a good one, but I guess I'm really not seeing Windows support being a first-class requirement when a majority of servers are ran on linux boxes.
Also FWIW, I believe tauri uses Tokio, and they aim to be the cross-platform experience for apps using TS+Rust.
My 2¢ after spending a month debugging io_uring issues in my project: I'd avoid using io_uring unless it's absolutely necessary. It gets really messy in many scenarios.
Also, regarding monoio - our experience wasn't great. The codebase was problematic, with frequent crashes in production and unnecessary heap allocations during vectored writes.
Describe the problem related to your feature request.
Compared with
tokio_uring
andmonoio
,compio
has better performance and compatibility, and has better support for the windows platform.What solution would you like?
Replace
tokio_uring
in valence and usecompio
What alternative(s) have you considered?
monoio
, but it does not support windows platformAdditional context
As far as I know, Apache OpenDAL™ is migrating to
compio
and there may be detailed performance testing there.The text was updated successfully, but these errors were encountered: