Updating code for e26 and c14

Starting with art 3.12, code can be compiled with e26 (GCC 12.1.0) and c14 (Clang 14.0.6). As expected, the newer compilers find more infelicities in the code, some of which indicate a need for code review.

Issues found by e26

• may be used uninitialized

  /cvmfs/larsoft.opensciencegrid.org/products/gcc/v12_1_0/Linux64bit+3.10-2.17/include/c++/12.1.0/bits/std_abs.h:76:31: error: 'dir_y_prime' may be used uninitialized [-Werror=maybe-uninitialized]
   76 |   { return __builtin_fabsf(__x); }
      |                               ^
  /home/garren/scratch/larsoft/uboone/dev5/srcs/ubana/ubana/searchingfornues/Selection/CommonDefs/Geometry.h In function 'std::vector<float, std::allocator<float> > searchingfornues::polarAngles(float, float, float, size_t, size_t)':
  /home/garren/scratch/larsoft/uboone/dev5/srcs/ubana/ubana/searchingfornues/Selection/CommonDefs/Geometry.h:110:11: note: 'dir_y_prime' was declared here
  110 |     float dir_y_prime, dir_z_prime;
      |           ^~~~~~~~~~~
  

  The code should be reviewed. However, a temporary workaround is to use -Wno-maybe-uninitialized.
  Note that -Wno-maybe-uninitialized is valid for GNU, not Clang. Use the following in cet_set_compiler_flags:

  $<$<CXX_COMPILER_ID:GNU>:-Wno-maybe-uninitialized>
  

• copies and references

  /home/garren/scratch/larsoft/uboone/dev5/srcs/ubana/ubana/Filters/NumuCCinclDumper/NumuCCinklRun3_module.cc: In member function 'void NumuCCana::fillMCInfo(const art::Event&)':
  /home/garren/scratch/larsoft/uboone/dev5/srcs/ubana/ubana/Filters/NumuCCinclDumper/NumuCCinklRun3_module.cc:814:25: error: loop variable 'evt_weight' creates a copy from type 'const std::pair<const std::__cxx11::basic_string<char>, std::vector<double, std::allocator<double> > >' [-Werror=range-loop-construct]
  814 |         for (auto const evt_weight : evtwgt_map){
      |                         ^~~~~~~~~~
  /home/garren/scratch/larsoft/uboone/dev5/srcs/ubana/ubana/Filters/NumuCCinclDumper/NumuCCinklRun3_module.cc:814:25: note: use reference type to prevent copying
  814 |         for (auto const evt_weight : evtwgt_map){
      |                         ^~~~~~~~~~
      |                         &
  

  Apply the suggested fix.

-        for (auto const evt_weight : evtwgt_map){
+        for (auto const& evt_weight : evtwgt_map){

• array bounds

  In member function 'virtual void ub::LowLevelNueFilter::analyze(const art::Event&)':
  /home/garren/scratch/larsoft/uboone/dev5/srcs/ubana/ubana/TPCNeutrinoIDFilter/LowLevelNueFilter_module.cc:1224:163: error: array subscript 1 is outside array bounds of 'double [1]' [-Werror=array-bounds]
   1224 |               std::cout<<"pfparticle reco pdgcode "<<pdgcode<<" associated vertex ID is"<<PrimaryVertexID<<" XYZ= ["<<PrimaryVertexXYZ[0]<<","<<PrimaryVertexXYZ[1]<<","<<PrimaryVertexXYZ[2]<<"] !!!!!"<<std::endl;
      |
  

  The code should be reviewed. However, a temporary fix is to use the -Wno-array-bounds flag. This flag is valid for GNU only.

• std::binary_function is deprecated

  In file included from /home/garren/scratch/larsoft/dune/dev5/srcs/dunereco/dunereco/FDSensOpt/IniSegAlg/IniSegAlg.cxx:9:
  /home/garren/scratch/larsoft/dune/dev5/srcs/dunereco/dunereco/FDSensOpt/IniSegAlg/IniSegAlg.h:97:21 error: ‘template<class _Arg1, class _Arg2, class _Result> struct std::binary_function’ is deprecated [-Werror=deprecated-declarations]
   97 |         public std::binary_function< Hit2D, Hit2D, bool >
      |                     ^~~~~~~~~~~~~~~
  

  The workaround is to add -Wno-deprecated-declarations to cet_set_compiler_flags. The code should be reviewed.

• error: ignoring return value
  • The code should be reviewed.
  • The workaround is to add -Wno-unused-result to cet_set_compiler_flags.

Issues found by c14

• variable set but not used
  • The code should be reviewed.
  • The unused variable may be commented out, but this may indicate code that is not working as intended.
• returning address of local temporary object

  /home/garren/scratch/larsoft/uboone/dev5/srcs/ubcrt/ubcrt/CRT/CRTBernFEBDAQCore/Overlays/BernZMQFragment.hh:95:74: error: returning address of local temporary object [-Werror,-Wreturn-stack-address]
  const char* c_str() const { std::ostringstream ss; ss << *this; return ss.str().c_str(); }
                                                                         ^~~~~~~~
  

  The code should be reviewed. A temporary workaround is to use -Wno-return-stack-address. This flag is only valid for Clang.

• wirecell prior to 0.25.1
  • versions of wirecell prior to 0.25.1 invoke _Complex if the Persist.h header is used.
  • c14 complains that '_Complex' is a C99 extension
  • When building with wirecell prior to 0.25.1, include $<$<CXX_COMPILER_ID:Clang>:-Wno-c99-extensions> in the call to cet_set_compiler_flags.
  • This problem is fixed in wirecell 0.25.1
• object backing the pointer will be destroyed at the end of the full-expression
  • The code should be reviewed.
• c14: struct is non C compatible for linkage
  • The code should be reviewed.
  • Problem related to void init() member
• implicit conversion changes value
  • The code should be reviewed.
  • Note that c14 will accept -Wno-implicit-const-int-float-conversion, but c7 does not recognize this flag.

  /home/garren/scratch/larsoft/uboone/dev5/srcs/ubana/ubana/searchingfornues/Selection/AnalysisTools/ContainmentAnalysis_tool.cc:304:11: error: implicit conversion from 'std::numeric_limits<int>::type' (aka 'int') to 'float' changes value from 2147483647 to 2147483648 [-Werror,-Wimplicit-const-int-float-conversion]
  _dvtx = std::numeric_limits<int>::max();
        ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

castxml

• srproxy v00.43 with castxml v0_5_1 is required.
  • See redmine issue 28198
  • e26 also requires an additional -fsized-deallocation in the gen_srproxy call

- COMMAND gen_srproxy --flat -i sbnanaobj/StandardRecord/StandardRecord.h -o FlatRecord --target caf::StandardRecord --include-path $ENV{SBNANAOBJ_DIR}:$ENV{ROOT_INC} -op sbnanaobj/StandardRecord/Flat/ --prolog $ENV{SBNANAOBJ_DIR}/sbnanaobj/StandardRecord/Flat/Prolog.h --extra-cflags ' -D_Float16=short'
+ COMMAND gen_srproxy --flat -i sbnanaobj/StandardRecord/StandardRecord.h -o FlatRecord --target caf::StandardRecord --include-path $ENV{SBNANAOBJ_DIR}:$ENV{ROOT_INC} -op sbnanaobj/StandardRecord/Flat/ --prolog $ENV{SBNANAOBJ_DIR}/sbnanaobj/StandardRecord/Flat/Prolog.h --extra-cflags ' -D_Float16=short -fsized-deallocation'