The goal is to allow the LSP bindings of features to be defined outside
the ClangdLSPServer class, turning it into less of a monolith.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
whoops, meant to create a new review for that followup change, not update this one...
LGTM, thanks! (I wish outgoing calls were not so different :/)
clang-tools-extra/clangd/ClangdLSPServer.cpp | ||
---|---|---|
173–180 | why not keep the old lookup style ? since handlers are unique_functions, checking for null should still suffice (and be cheap) | |
clang-tools-extra/clangd/LSPBinder.h | ||
68 | s/peek/load | |
clang-tools-extra/clangd/unittests/LSPBinderTests.cpp | ||
19 | please fix | |
21 | please fix | |
26 | please fix (we usually follow camelCase for json-serializable types in clangd, but i don't think it is worth doing in tests) | |
29 | nit: s/P/P.field("x") | |
52 | please fix, and maybe have a counter, in addition to last value to check for "invalid type" case? | |
70 | nit: ASSERT_THAT(Reply.hasValue()) here and elsewhere |
why not keep the old lookup style ? since handlers are unique_functions, checking for null should still suffice (and be cheap)