yellowhatter

@Mallets tagging you here

Everything is currently here #596

Here is a test. ``` #[test] fn session_perf() { let many_sessions: Vec = (0..10).map(|_| zenoh::open(zenoh::config::peer()).res_sync().unwrap() ).collect(); assert!(many_sessions.len() == 10); } ``` It takes 8 seconds to run it on my...

What about pre-initializing discovery data from already existing sessions when creating a new one? - if this would be possible, we could also avoid awaiting for `scouting_delay`.

> @yellowhatter could you verify if this issue is still relevant with the rework SHM being done in https://github.com/eclipse-zenoh/zenoh/tree/dev/1.0.0? Ok, I'll do.

@Mallets yes, this issue is old and outdated

@Mallets @milyin @sashacmc @DenisBiryukov91 This is a draft for SHM C API

Hello, @MelamudMichael ! Can you please describe your testing scenario to achieve this?

Hello, @MelamudMichael ! In actual release filesystem-mapped shm segments are really not erased in case of owner process termination. The new SHM codebase in the upcoming release will have this...

I think we need to investigate the pipe spec on it's behavior when other pipe end closes. That might be the reason of this bug.