VSCode rust-analyzer panics with index out of bounds when toolchain missing

Two worker threads crashed while contributors were editing the rust-analyzer repository: "thread 'Worker9' (3966083) panicked at crates/rust-analyzer/src/handlers/notification.rs:338:25: index out of bounds: the len is 0 but the index is 0" and "thread 'Worker14' (3966088) panicked at crates/rust-analyzer/src/handlers/notification.rs:338:25: index out of bounds: the len is 0 but the index is 0." The failure site is precise: crates/rust-analyzer/src/handlers/notification.rs at column 25 of line 338 in commit 4dba1b4, with the rust-analyzer binary identified as "rust-analyzer version: 0.4.2791-standalone [/home/gh-bjorn3/.vscode-server/extensions/rust-lang.rust-analyzer-0.4.2791-linux-arm64/server/rust-analyzer]."

Logs include truncated backtraces showing the unwind path up to the bounds check: "0: __rustc::rust_begin_unwind 1: core::panicking::panic_fmt 2: core::panicking::panic_bounds_check 3: core::ops::function::FnOnce::call_once{{vtable.shim}} 4: <alloc::boxed::Box<F,A> as core::ops::function::FnOnce<Args>>::call_once 5: core::ops::function::FnOnce::call_once{{vtable.shim}} 6: <alloc::boxed::Box<F,A> as core::ops::function::FnOnce<Args>>::call_once" and the logs advise "Note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace." Timestamps tied to the errors are 2026-02-13T10:39:40.26494485Z and 2026-02-13T10:39:44.844284722Z, which bracket the sequence of flycheck and panic messages.

The report surfaces flycheck as a likely trigger point: a log line shows "world.flycheck.restart_workspace(saved_file);" followed by "2026-02-13T10:39:40.26494485Z ERROR flycheck task panicked: Any { .. }" and later "2026-02-13T10:39:44.844284722Z ERROR should have exactly one flycheck handle when invocation strategy is once." The environment was VSCode using the linux-arm64 extension build, and the runtime state included "rustc version: Rust toolchain is uninstalled." The reporter explicitly noted, "Rust-analyzer panics at some point when working on the rust-analyzer repo. Probably related to the fact that the active toolchain was not installed when rust-analyzer started." The config is described as "Fairly standard config for the rust repo."

Key missing items that would prove causality are absent from the logs: the exact code around line 338 that performs the indexing operation is not included, the full RUST_BACKTRACE output was not captured, and the reproduction steps referenced in the original report are not present in the supplied materials. The backtrace recommendation is actionable: reproduce the run with RUST_BACKTRACE=full to capture the call chain beyond the bounds check and to reveal which collection or index triggered the panic.

To help triage, capture and attach the full RUST_BACKTRACE=full output for Worker9 and Worker14, include the rust-analyzer binary path and commit id 4dba1b4, and test whether the panic occurs only when "Rust toolchain is uninstalled." Inspect crates/rust-analyzer/src/handlers/notification.rs around line 338 for unchecked indexing, and compare flycheck handle management code around "world.flycheck.restart_workspace(saved_file);" to the log "should have exactly one flycheck handle when invocation strategy is once." Those artifacts - full backtrace, binary path, commit hash, timestamps, and thread ids - will give maintainers the concrete traces needed to fix the index-out-of-bounds in v0.4.2791-standalone.