When loading workspaces fails, tell the user if the `linkedProjects` config is set

[Veykril]

Currently when someone has the linkedProjects config set and their workspace fails to load (due to the config pointing to non-existing paths or similar), the user will just get confused as r-a doesn't tell them anything else. We should tell them loud and clear that it is most likely a misconfiguration of the linkedProjects config causing them problems.

[stupendoussuperpowers]

could you clarify what you expect the error message to be/what you see currently?

for me, I see the following when it fails to load a linkedProject from the settings, which does point to a failure and a suggestion with it -

[ERROR rust_analyzer::config] failed to load linked project: project root must point to Cargo.toml or rust-project.json: /home/../../.\Cargo.toml

[Veykril]

Relevant part to check is here

rust-analyzer/crates/rust-analyzer/src/reload.rs

Lines 175 to 178 in 8a23314

	if let Err(_) = self.fetch_workspace_error() {
	status.health = lsp_ext::Health::Error;
	message.push_str("Failed to load workspaces.\n\n");
	}

We should check whether the linkedProjects config is set, and if yes add a note to that message that the user should check whether those projects are valid/exist.

[stupendoussuperpowers]

will submit a pull request for this.

[bionicles]

currently suffering from this .. hidden config somewhere I never set up, pointing to folders I moved, and there's no indication as to where the "linkedProjects" is, just a "failed to load workspace" except I don't even want to use a workspace, i just want rust-analyzer to work.

global state is bad enough, but when it's causing an error, to not notify the user which global state config file is the source of the cached out of date paths, that's just dumb

steps to reproduce: no idea

[ERROR rust_analyzer::main_loop] FetchWorkspaceError:
rust-analyzer failed to load workspace: Failed to load the project at /home/bion/hax/etl/Cargo.toml: Failed to query rust toolchain version at /home/bion/hax/etl, is your toolchain setup correctly?: cd "/home/bion/hax/etl" && "cargo" "--version" failed: No such file or directory (os error 2)

i moved the "etl" crate inside the "etl_warehouse" cargo workspace, but vscode keeps trying to load the old path, i've deleted every .vscode folder i can find, it's not in settings.json, i deleted the .vscode-server user history, still there is some source of stale global mutable state i don't want, never set up myself, but vscode won't get over it and just ... go to the path cargo.toml i'm looking at right now. guess i'll delete ".vscode-server" too, nice, love it