Figure out what the team is willing/unwilling to tolerate re. external CI or CI setups involving external components for rust-lang/rust and other rust-lang/* repos #201
Description
Recently, some question about "what are we willing/unwilling to tolerate for CI setups involving external pieces" have come up, for:
- The IBM-provided runner setup: Investigate possibility of requesting IBM runner access for {ppc64le, s390x} #199
- Where the team had concerns regarding the GitHub App requiring admin access for rust-lang/rust, but was okay with rust-lang/{compiler-builtins,libc,stdarch}.
- Community groups wishing to pursue higher target support tiers (notably Tier 1) for e.g. RISC-V. For instance RFC: Promote riscv64gc-unknown-linux-gnu to Tier-1 (without host tools) rfcs#3707 and #t-infra > We're eager to provide the community with RISC-V CI infrastr.
- One group had a concrete infra setup proposal #t-infra > We're eager to provide the community with RISC-V CI infrastr @ 💬, but this is hard to evaluate without the team figuring out what we are willing/unwilling to tolerate.
- For cases like the RISC-V proposal, the infra team was also interested to speak to the Foundation about whether the team would feel safer if there are more formal arrangements between the Foundation / External Providers re. longer-term availability and such.