Does it make sense to have all the "tidy checks" be named after the coding
style they originate from? I imagine that many will be shared. E.g., instead of
`google-explicit-constructor`, it seems like just calling it
`explicit-constructor` would be better, since this seems like a fairly general
check that a lot of people would want. Then each "coding standard" would mostly
just be a set of checks, similar to how clang-format styles are just a group of
style settings.
Some settings are truly project specific, like LLVM's include order (which
depends on LLVM's directory layout). For those, I think it makes sense to have
them prefixed with the project/coding standard name.
================
Comment at: clang-tidy/llvm/LLVMModule.cpp:88
@@ +87,3 @@
+ StringRef SourceFile = Sources.getFilename(HashLoc);
+ if (!SourceFile.endswith(".cc"))
+ return;
----------------
Shouldn't this be checking LangOptions instead of the extension? e.g. this
wouldn't work for LLVM's `.cpp` naming.
================
Comment at: test/clang-tidy/select-checks.cpp:2
@@ +1,3 @@
+// RUN: grep -Ev "// *[A-Z-]+:" %s > %t.cpp
+// RUN: clang-tidy %t.cpp -fix -checks=llvm.* --
+// RUN: FileCheck -input-file=%t.cpp %s
----------------
Why is this `llvm.` when the check names are `llvm-`? Seems confusing.
http://llvm-reviews.chandlerc.com/D884
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
