Jim Newsome

I'm still using my own replacement for `archived_root` to avoid the miri stacked borrows failure. Maybe this or something like it should be added to rkyv: ``` /// Taking the...

Yeah, seems like there's no quick answer, other than documenting or removing it in the meantime. I wonder if the `SerializeWith` and `Serialize` traits could be modified to specify that...

Tor discussion on whether to recommend BBR for relay operators: https://gitlab.torproject.org/tpo/core/tor/-/issues/30420

Shadow uses its own implementation of TCP, so it doesn't really matter whether the kernel Shadow is running on has BBR support enabled. The bulk of the complexity+work is faithfully...

> In the case of fifos (see man 7 fifo), it can lead to deadlock. This part overlaps a bit with #1706 and #1723. It may or may not make...

Alternatively we could *never* migrate workers across numa nodes. That again could perform badly if the initial assignment is too unbalanced across numa nodes, but is probably the simplest option...

Related: https://github.com/shadow/shadow/issues/1250

My initial attempt looks like this: ``` guix environment \ --pure \ --ad-hoc \ cmake \ clang-toolchain \ make \ pkg-config \ igraph \ glib \ bash \ python \...

Btw we might consider supporting a [nix](https://en.wikipedia.org/wiki/Nix_(package_manager)) build for similar reasons. They're pretty similar in most respects that matter; I currently slightly prefer guix, mostly becauses it uses guile scheme...

As noted [previously](https://github.com/shadow/shadow/issues/1839#issuecomment-1006103771), managed programs in a guix-built shadow simulation will probably also need to be built with guix unless we remove the shim's runtime dependency on libc (which is...