Rework cmd tests

[alexheretic]

• Rewrite all tests/tests.rs "cmd" tests
• cargo update
• cmd test wait-for-rls-stdout timeout reduced 300 -> 30

The cmd tests aren't good enough because they don't fail very usefully. They either work or block for 300 seconds then say nothing. For example the build is currently failing in master, but can anyone see why?

The new code has

• Explicit timeouts for any blocking calls, meaning it's easy to see where we wait for stdout.
• More accurate dynamic json assertions.

  assert_eq!(json["method"], "window/progress");
  assert_eq!(json["params"]["title"], "Indexing");

• Removed all but one test checking window/progress messages.
• Removed outer timeout/closure which obfuscates test panics & adds complexity.

The RlsHandle now uses a thread to eagerly read stdout, and eprints that stdout when an assertion fails. This stdout allows us to debug the tests when they fail. To see the improvement here is the failure output of the tests with this change:

running 4 tests
test cmd_test_infer_bin ... FAILED
test cmd_test_complete_self_crate_name ... ok
test cmd_test_simple_workspace ... ok
test cmd_changing_workspace_lib_retains_bin_diagnostics ... ok

failures:

---- cmd_test_infer_bin stdout ----
thread 'cmd_test_infer_bin' panicked at 'assertion failed: `(left == right)`
  left: `String("window/showMessage")`,
 right: `"textDocument/publishDiagnostics"`', tests/tests.rs:58:5
note: Run with `RUST_BACKTRACE=1` for a backtrace.
---rls-stdout---
Content-Length: 607

{"jsonrpc":"2.0","id":0,"result":{"capabilities":{"textDocumentSync":2,"hoverProvider":true,"completionProvider":{"resolveProvider":true,"triggerCharacters":[".",":"]},"definitionProvider":true,"implementationProvider":true,"referencesProvider":true,"documentHighlightProvider":true,"documentSymbolProvider":true,"workspaceSymbolProvider":true,"codeActionProvider":true,"codeLensProvider":{"resolveProvider":false},"documentFormattingProvider":true,"documentRangeFormattingProvider":false,"renameProvider":true,"executeCommandProvider":{"commands":["rls.applySuggestion-29146","rls.deglobImports-29146"]}}}}Content-Length: 92

{"jsonrpc":"2.0","method":"window/progress","params":{"id":"progress_1","title":"Building"}}Content-Length: 104

{"jsonrpc":"2.0","method":"window/progress","params":{"done":true,"id":"progress_1","title":"Building"}}Content-Length: 92

{"jsonrpc":"2.0","method":"window/progress","params":{"id":"progress_0","title":"Indexing"}}Content-Length: 157

{"jsonrpc":"2.0","method":"window/showMessage","params":{"message":"Cargo failed: failed to parse manifest at `/home/alex/project/rls/Cargo.toml`","type":1}}Content-Length: 104

{"jsonrpc":"2.0","method":"window/progress","params":{"done":true,"id":"progress_0","title":"Indexing"}}
---------------


failures:
    cmd_test_infer_bin

test result: FAILED. 3 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out

Once you have this output it's fairly easy to see Cargo failed: failed to parse manifest at /home/alex/project/rls/Cargo.toml. The recent update (#1058) to the main Cargo.toml is was interfering with the cmd tests.

And all this with ~100 lines net removed :)

[Xanewok]

I've been meaning to tackle the recent infer_bin failure myself, thank you! It's a lot better to see what's causing the actual failure. Could you rebase after the #1063 merge, please?

[Xanewok]

cc @spacekookie who also wanted to tackle the integration "cmd" tests and do more structural checking

[Xanewok]

Hey @alexheretic, I'm sorry for the churn - could you please rebase it again and only include the testing changes (it seems that cargo update was done recently already) to mitigate potential conflicts? Thanks! I'd love to land this and include more tests.

[alexheretic]

np

[Xanewok]

Thanks once again!

bors r+

[bors-voyager]

Build succeeded

• continuous-integration/travis-ci/push
• continuous-integration/appveyor/branch