The idea here is to unify the different ways in which we can currently set options on SourceKit-LSP in a scalable way: Environment variables, command line arguments to `sourcekit-lsp` and initialization options.
The idea is that a user can define a `~/.sourcekit-lsp/.sourcekit-lsp` file (we store logs in `~/.sourcekit-lsp/logs` on non-Darwin platforms), which will be used as the default configuration for all SourceKit-LSP instances. They can also place a `.sourcekit-lsp` file in the root of a workspace to configure SourceKit-LSP for that project specifically, eg. setting arguments that need to be passed to `swift build` for that project and which thus also need to be set on SourceKit-LSP.
For compatibility reasons, I’m mapping the existing command line options into the new options structure for now. I hope to delete the command line arguments in the future and solely rely on `.sourcekit-lsp` configuration files.
Environment variable will be migrated to `.sourcekit-lsp` in a follow-up commit.
# Conflicts:
# Sources/SourceKitLSP/SourceKitLSPServer+Options.swift
# Sources/SourceKitLSP/Swift/SwiftLanguageService.swift
# Sources/sourcekit-lsp/SourceKitLSP.swift
# Tests/SourceKitLSPTests/BackgroundIndexingTests.swift
# Tests/SourceKitLSPTests/ExecuteCommandTests.swift
- Rename methods to highlight that we’re talking about generated interfaces here, not `.swiftinterface` files
- Don’t open the generated interface in `documentManager`. Opening documents in `documentManager` should only be done by the `textDocument/didOpen` notification from the LSP client. Otherwise we might indefinitely keep the document in the document manager
- After getting the generated interface from sourcekitd, close the document in sourcekitd again. We don’t provide semantic functionality in the generated interface yet, so we can’t interact with the generated interface path. Before, we left it open in sourcekitd indefinitely.
- A couple of code simplifications.
Fixes#878
rdar://116705653
This was causing a non-deterministic test failure: When target preparation finishes while a diagnostic request is in progress, it will re-open the document, which calls `DiagnosticReportManager.removeItemsFromCache` for that document’s URI. With the old implementation, we would thus cancel the diagnostics sourcekitd request and return a cancelled error to the diagnostics LSP request.
While doing this, I also realized that there was a race condition: Document re-opening would happen outside of the SourceKit-LSP message handling queue and could thus run concurrently to any other request. This means that a sourcekitd request could run after `reopenDocument` had closed the document but before it was opened again. Introduce an internal reopen request that can be handled on the main message handling queue and thus doesn’t have this problem
There were two issues with Objective-C XCTest discovery:
1. We were relying on syntactic test discovery after a document is edited. But since we don't have syntactic test discovery for Objective-C tests, this meant that all tests would disappear as a document got edited. Until we get syntactic test discovery for Objective-C, use the semantic index to discover tests, even if they are out-of-date.
2. We were assuming that the `DocumentSymbols` request returned `[DocumentSymbol]` to find the ranges of tests. But clangd returns the alternative `[SymbolInformation]`, which meant that we were only returning the position of a test function’s name instead of the test function’s range.
rdar://126810202
When the semantic index is out-of-date, we currently purely rely on the syntactic index to discover tests and completely ignore data from the semantic index. This may lead to confusing behavior. For example if you have
```
class MightInheritFromXCTestCaseOrNot {}
class MyClass: MightInheritFromXCTestCaseOrNot {
func testStuff() {}
}
```
Then we don’t return any tests when the semantic index is up-to-date. But once the file is modified (either on disk or in-memory), we purely rely on the syntactic index, which reports `testStuff` as a test method. After a build / background indexing finishes, the test method disappears again.
We can mitigate this problem as follows: If we have stale semantic index data for the test file, for every test method found by the syntactic index, check if we have an entry for this method in the semantic index. If we do, but that entry is not marked as a test class/method, we know that the semantic index knows about this method but decided that it’s not a test method for some reason. So we should ignore it.
rdar://126492948
If the index for a given file is not up-to-date, perform a syntactic scan for tests within it.
This should allow editors to run the `textDocument/tests` request while the user is editing a file to scan it for test cases.
The naming was quite inconsistent here. Let’s rename these to `LanguageService` to highlight that they belong together.
- ToolchainLanguageServer -> LanguageService
- SwiftLanguageServer -> SwiftLanguageService
- ClangLanguageServerShim -> ClangLanguageService
