Issue #958: Travis CI should enforce clang-format standards

[baylesj]

This patch adds clang format support to the travis bots.

[baylesj]

Any issues with pulling the run-clang-format.py script? It's MIT licensed.

[dota17]

Some questions for me.

1. After running these command from Building

cd jsoncpp/
BUILD_TYPE=release
#plain, debug, debugoptimized, release, minsize
LIB_TYPE=shared
#LIB_TYPE=static
meson --buildtype ${BUILD_TYPE} --default-library ${LIB_TYPE} . build-${LIB_TYPE}
ninja -v -C build-${LIB_TYPE} test

I run

ninja -v -C build-${LIB_TYPE} clang-format

and I got these changes (some changes were wrong) :
run-clang-format.txt

2. From PR Just run clang format #1025 , I found that
   Value& root and IStream & is, Value & root
   before char &, some places have a blank, some places have no.

3. To test this PR, I run some tests by pushing some wrong format codes. It is OK to check the wrong format.
   But the third test [Mac clang meson static release testing] is error, and the error message is
   run-clang-format.py: error: Command 'clang-format --version' failed to start: [Errno 2] No such file or directory
   Is it better to remove format check in third test or update the env?

[baylesj]

1. This should not change the results of ninja clang-format at all, so I'm not sure what the cause of what you are seeing is. This only affects Travis. Update: I have changed our clang format settings so that it is more consistent and should look better now.

2. Updated to fix pointer binding. They renamed the config value.

3. I can fix that, I think clang format isn't installing properly on mac.

[dota17]

hi, I have checked the new commit and find the question 2 and 3 have been repaired.
however, as for the question 1,I have some questions.

1. After merging the fix-deprecated to clang-formatize and building, I get this result:
   test.txt

+[[noreturn]] void throwRuntimeError(String const& msg) {
+  abort();
+}[[noreturn]] void throwLogicError(String const& msg) {
+  abort();
+}

this change is wrong and it means this clang-format standards still have some questons.

2. the clang-format result in local is different from the result of Travis CI.
   I run ./travis_scripts/run-clang-format.sh in my local computer and get the same changes with test.txt.

run-clang-format.sh: 3: run-clang-format.sh: Bad substitution
--- /root/jsoncpp/community/jsoncpp/.travis_scripts/../src/lib_json/json_value.cpp      (original)
+++ /root/jsoncpp/community/jsoncpp/.travis_scripts/../src/lib_json/json_value.cpp      (reformatted)
@@ -206,14 +206,17 @@
 Exception::~Exception() JSONCPP_NOEXCEPT {}
 char const* Exception::what() const JSONCPP_NOEXCEPT { return msg_.c_str(); }
 RuntimeError::RuntimeError(String const& msg) : Exception(msg) {}
-LogicError::LogicError(String const& msg) : Exception(msg) {}
-[[noreturn]] void throwRuntimeError(String const& msg) {
+LogicError::LogicError(String const& msg)
+    : Exception(msg){}[[noreturn]] void throwRuntimeError(String const& msg) {
   throw RuntimeError(msg);
 }
 [[noreturn]] void throwLogicError(String const& msg) { throw LogicError(msg); }
 #else // !JSON_USE_EXCEPTION
-[[noreturn]] void throwRuntimeError(String const& msg) { abort(); }
-[[noreturn]] void throwLogicError(String const& msg) { abort(); }
+[[noreturn]] void throwRuntimeError(String const& msg) {
+  abort();
+}[[noreturn]] void throwLogicError(String const& msg) {
+  abort();
+}
 #endif

 // //////////////////////////////////////////////////////////////////
@@ -227,7 +230,9 @@
 // Notes: policy_ indicates if the string was allocated when
 // a string is stored.

-Value::CZString::CZString(ArrayIndex index) : cstr_(nullptr), index_(index) {}
+Value::CZString::CZString(ArrayIndex index)
+    : cstr_(nullptr), index_(index) {
+}

 Value::CZString::CZString(char const* str, unsigned length,
                           DuplicationPolicy allocate)

The results are same.

[dota17]

The Travis CI checks were not successful, Fixed it by #1036

[dota17]

The above problem in my local env had been solved.
Tips: use the current version - clang-8 and clang-format-8

[cdunn2001]

Very nice!

[cdunn2001]

Hmmm. Today I hit a tiny clang-format error. I guess we have to run clang-format now locally before pushing. I've added a script for that in #1229. I hope different versions of clang-format don't disagree with each other.