Release/2.7.5

[jholloc]

Addressing issues with SSL. Also adding running of UDA integration tests to CI, including running tests with SSL authenticated client/server.

Fixing issues:

• IMAS-5005
• IMAS-4862

[jholloc]

@SimonPinches just pushed the same fix for server2.

[SimonPinches]

Thanks @jholloc - all works now :-)

[SimonPinches]

Compilation fails with Intel compilers (see IMAS-5014):

[130/337] Building CXX object source/client2/CMakeFiles/client2-objects.dir/client.cpp.o
FAILED: source/client2/CMakeFiles/client2-objects.dir/client.cpp.o
/work/imas/opt/EasyBuild/software/iccifort/2020.4.304/compilers_and_libraries_2020.4.304/linux/bin/intel64/icpc -DA64 -DFMT_LOCALE -DLIBXML2_PRINTF_CHAR_ARG -DNOLIBMEMCACHED -DNOTGENERICENABLED -DNO_GSL_LIB -DOSTYPE=Linux -DUDA_EXPORT -D__TIRPC__ -I/opt/local/include -Isource -I/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source -I/work/imas/opt/EasyBuild/software/OpenSSL/1.1/include -I/work/imas/opt/EasyBuild/software/libxml2/2.9.10-GCCcore-10.2.0/include/libxml2 -I/include -I/work/imas/opt/EasyBuild/software/libtirpc/1.3.1-GCCcore-10.2.0/include/tirpc -isystem /work/imas/opt/EasyBuild/software/Boost/1.74.0-iccifort-2020.4.304/include -isystem /work/imas/opt/EasyBuild/software/fmt/7.1.1-GCCcore-10.2.0/include -O2 -axAVX,CORE-AVX2 -ftz -fp-speculation=safe -fp-model source -fPIC -std=c++17 -fPIC -pthread -Wno-use-after-free -Wno-mismatched-new-delete -g -fPIC   -Wall -Werror -Wextra -Wno-unused-parameter -Wno-missing-field-initializers -MD -MT source/client2/CMakeFiles/client2-objects.dir/client.cpp.o -MF source/client2/CMakeFiles/client2-objects.dir/client.cpp.o.d -o source/client2/CMakeFiles/client2-objects.dir/client.cpp.o -c /tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/client2/client.cpp
icpc: command line warning #10148: option '-Wno-use-after-free' not supported
icpc: command line warning #10148: option '-Wno-mismatched-new-delete' not supported
/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/client2/client.cpp(586): error #1595: non-POD (Plain Old Data) class type passed through ellipsis
              UDA_LOG(UDA_LOG_ERROR, "Error identifying the Data Source [%s]\n", data_source);
              ^

compilation aborted for /tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/client2/client.cpp (code 2)

[SimonPinches]

Can we fix to also build with Intel compilers?

[SimonPinches]

Still have problems compiling with Intel:

FAILED: source/bin/CMakeFiles/uda_cli.dir/uda_cli.cpp.o
/work/imas/opt/EasyBuild/software/iccifort/2020.4.304/compilers_and_libraries_2020.4.304/linux/bin/intel64/icpc -DA64 -DCAPNP_ENABLED=1 -DFM
T_LOCALE -DLIBXML2_PRINTF_CHAR_ARG -DOSTYPE=Linux -DUDA_EXPORT -I/opt/local/include -Isource -I/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-202
0.4.304/UDA-release-2.7.5/source -I/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/extlib/GSL-4.0.0/include -isystem
/work/imas/opt/EasyBuild/software/Boost/1.74.0-iccifort-2020.4.304/include -isystem /work/imas/opt/EasyBuild/software/fmt/7.1.1-GCCcore-10.2
.0/include -O2 -axAVX,CORE-AVX2 -ftz -fp-speculation=safe -fp-model source -fPIC -std=c++17 -fPIC -pthread -Wno-use-after-free -Wno-mismatch
ed-new-delete -g -fPIE   -Wall -Werror -Wextra -Wno-unused-parameter -Wno-missing-field-initializers -MD -MT source/bin/CMakeFiles/uda_cli.d
ir/uda_cli.cpp.o -MF source/bin/CMakeFiles/uda_cli.dir/uda_cli.cpp.o.d -o source/bin/CMakeFiles/uda_cli.dir/uda_cli.cpp.o -c /tmp/pinches/Ea
syBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/bin/uda_cli.cpp
icpc: command line warning #10148: option '-Wno-use-after-free' not supported
icpc: command line warning #10148: option '-Wno-mismatched-new-delete' not supported
/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/extlib/GSL-4.0.0/include/gsl/span(128): error #597: "gsl::details::sp
an_iterator<Type>::operator gsl::details::span_iterator<const char>() const noexcept [with Type=const char]" will not be called for implicit
 or explicit conversions
          constexpr operator span_iterator<const Type>() const noexcept
                    ^
          detected during:
            instantiation of class "gsl::details::span_iterator<Type> [with Type=const char]" at line 633
            instantiation of "gsl::span<ElementType, Extent>::iterator gsl::span<ElementType, Extent>::begin() const noexcept [with ElementT
ype=const char, Extent=18446744073709551615UL]" at line 24 of "/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source
/bin/uda_cli.cpp"
            instantiation of "std::ostream &operator<<(std::ostream &, gsl::span<T, 18446744073709551615UL>) [with T=const char]" at line 41
 of "/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/bin/uda_cli.cpp"
            instantiation of "std::ostream &operator<<(std::ostream &, const std::vector<T, std::allocator<T>> &) [with T=char]" at line 249
 of "/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/bin/uda_cli.cpp"

/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/extlib/GSL-4.0.0/include/gsl/span(128): error #597: "gsl::details::sp
an_iterator<Type>::operator gsl::details::span_iterator<const short>() const noexcept [with Type=const short]" will not be called for implic
it or explicit conversions
          constexpr operator span_iterator<const Type>() const noexcept
                    ^
          detected during:
            instantiation of class "gsl::details::span_iterator<Type> [with Type=const short]" at line 633
            instantiation of "gsl::span<ElementType, Extent>::iterator gsl::span<ElementType, Extent>::begin() const noexcept [with ElementT
ype=const int16_t={__int16_t={signed short}}, Extent=18446744073709551615UL]" at line 24 of "/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.
4.304/UDA-release-2.7.5/source/bin/uda_cli.cpp"
            instantiation of "std::ostream &operator<<(std::ostream &, gsl::span<T, 18446744073709551615UL>) [with T=const int16_t={__int16_
t={signed short}}]" at line 41 of "/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/bin/uda_cli.cpp"
            instantiation of "std::ostream &operator<<(std::ostream &, const std::vector<T, std::allocator<T>> &) [with T=int16_t={__int16_t
={signed short}}]" at line 252 of "/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/bin/uda_cli.cpp"

/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/extlib/GSL-4.0.0/include/gsl/span(128): error #597: "gsl::details::sp
an_iterator<Type>::operator gsl::details::span_iterator<const int>() const noexcept [with Type=const int]" will not be called for implicit o
r explicit conversions
          constexpr operator span_iterator<const Type>() const noexcept
                    ^
          detected during:
            instantiation of class "gsl::details::span_iterator<Type> [with Type=const int]" at line 633
            instantiation of "gsl::span<ElementType, Extent>::iterator gsl::span<ElementType, Extent>::begin() const noexcept [with ElementType=const int, Extent=18446744073709551615UL]" at line 24 of "/tmp/pinches/EasyBuild/UDA/2.7.5/iccifort-2020.4.304/UDA-release-2.7.5/source/bin/uda_cli.cpp"
...

[jholloc]

Not sure what we can do about the latest issue... looks like the Intel compiler doesn't like the GSL library, which is out of our control. I think this library is currently only used in the UDA CLI, so we could just turn of CLI compilation for Intel.

[olivhoenen]

Not better with last commit :-(

/home/ITER/hoeneno/GIT/UDA/extlib/GSL-4.0.0/include/gsl/span(128): error #597: "gsl::details::span_iterator<Type>::operator gsl::details::span_iterator<const int64_t={__int64_t={signed long}}>() const noexcept [with Type=const int64_t={__int64_t={signed long}}]" will not be called for implicit or explicit conversions
          explicit constexpr operator span_iterator<const Type>() const noexcept
                             ^
          detected during:
            instantiation of class "gsl::details::span_iterator<Type> [with Type=const int64_t={__int64_t={signed long}}]" at line 633
            instantiation of "gsl::span<ElementType, Extent>::iterator gsl::span<ElementType, Extent>::begin() const noexcept [with ElementType=const int64_t={__int64_t={signed long}}, Extent=18446744073709551615UL]" at line 85 of "/home/ITER/hoeneno/GIT/UDA/source/serialisation/capnp_serialisation.cpp"
            instantiation of "std::ostream &operator<<(std::ostream &, gsl::span<T, 18446744073709551615UL>) [with T=const int64_t={__int64_t={signed long}}]" at line 120 of "/home/ITER/hoeneno/GIT/UDA/source/serialisation/capnp_serialisation.cpp"
            instantiation of "void print_data<T>(std::ostream &, TreeNode::Array::Reader &, const std::string &) [with T=int64_t={__int64_t={signed long}}]" at line 158 of "/home/ITER/hoeneno/GIT/UDA/source/serialisation/capnp_serialisation.cpp"

[olivhoenen]

@jholloc trying some tricks I got from https://community.intel.com/t5/Intel-C-Compiler/Error-operator-will-not-be-called-for-implicit-or-explicit/td-p/1025378 it seems that I can continue the compilation with icpc, but my C++ template skills are not that accurate so I'll let you judge the pertinence and apply the patch if you think this is harmless:

diff --git a/extlib/GSL-4.0.0/include/gsl/span b/extlib/GSL-4.0.0/include/gsl/span
index ed90ecad..8e0df75d 100644
--- a/extlib/GSL-4.0.0/include/gsl/span
+++ b/extlib/GSL-4.0.0/include/gsl/span
@@ -125,6 +125,7 @@ namespace details
             : begin_(begin), end_(end), current_(current)
         {}

+       template <class Type>
         explicit constexpr operator span_iterator<const Type>() const noexcept
         {
             return {begin_, end_, current_};

I also have later another compilation error:

/home/ITER/hoeneno/GIT/UDA/source/plugins/template/templatePlugin.cpp(212): error #68: integer conversion resulted in a change of sign
      size_t narray = -1;
                      ^

compilation aborted for /home/ITER/hoeneno/GIT/UDA/source/plugins/template/templatePlugin.cpp (code 2)

[SimonPinches]

Thanks @jholloc , now I get the next error (but this one looks easier to fix):

[1/337] Building CXX object source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o
FAILED: source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o
/work/imas/opt/EasyBuild/software/intel-compilers/2023.2.1/compiler/2023.2.1/linux/bin/icpx -DA64 -DNOLIBMEMCACHED -DOSTYPE=Linux -DUDA_EXPORT -D__TIRPC__ -I/opt/local/include -I/tmp/pinches/EasyBuild/UDA/2.7.5/intel-compilers-2023.2.1/easybuild_obj/source -I/work/imas/opt/EasyBuild/software/OpenSSL/1.1/include -I/tmp/pinches/EasyBuild/UDA/2.7.5/intel-compilers-2023.2.1/UDA-release-2.7.5/source -I/work/imas/opt/EasyBuild/software/libxml2/2.11.5-GCCcore-13.2.0/include/libxml2 -I/work/imas/opt/EasyBuild/software/libtirpc/1.3.4-GCCcore-13.2.0/include/tirpc -isystem /work/imas/opt/EasyBuild/software/Boost/1.83.0-intel-compilers-2023.2.1/include -isystem /work/imas/opt/EasyBuild/software/fmt/10.2.0-GCCcore-13.2.0/include -O2 -axAVX,CORE-AVX2 -ftz -fp-speculation=safe -fp-model precise -fPIC -fPIC -pthread -Wno-use-after-free -Wno-mismatched-new-delete -g -std=c++17 -fPIC   -Wall -Werror -Wextra -Wno-unused-parameter -Wno-missing-field-initializers -MD -MT source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o -MF source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o.d -o source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o -c /tmp/pinches/EasyBuild/UDA/2.7.5/intel-compilers-2023.2.1/UDA-release-2.7.5/source/cache/memcache.cpp
error: unknown warning option '-Wno-use-after-free' [-Werror,-Wunknown-warning-option]

[jholloc]

@SimonPinches I've hopefully removed those warnings for the Intel compiler.

[SimonPinches]

Still fails with same error :-(

[1/331] Building CXX object source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o
FAILED: source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o 
/work/imas/opt/EasyBuild/software/intel-compilers/2023.2.1/compiler/2023.2.1/linux/bin/icpx -DA64 -DNOLIBMEMCACHED -DOSTYPE=Linux -DUDA_EXPORT -D__TIRPC__ -I/opt/local/include -I/tmp/pinches/EasyBuild/UDA/2.7.5/intel-compilers-2023.2.1/easybuild_obj/source -I/work/imas/opt/EasyBuild/software/OpenSSL/1.1/include -I/tmp/pinches/EasyBuild/UDA/2.7.5/intel-compilers-2023.2.1/UDA-release-2.7.5/source -I/work/imas/opt/EasyBuild/software/libxml2/2.11.5-GCCcore-13.2.0/include/libxml2 -I/work/imas/opt/EasyBuild/software/libtirpc/1.3.4-GCCcore-13.2.0/include/tirpc -isystem /work/imas/opt/EasyBuild/software/Boost/1.83.0-intel-compilers-2023.2.1/include -isystem /work/imas/opt/EasyBuild/software/fmt/10.2.0-GCCcore-13.2.0/include -O2 -axAVX,CORE-AVX2 -ftz -fp-speculation=safe -fp-model precise -fPIC -fPIC -pthread -Wno-use-after-free -Wno-mismatched-new-delete -g -std=c++17 -fPIC   -Wall -Werror -Wextra -Wno-unused-parameter -Wno-missing-field-initializers -MD -MT source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o -MF source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o.d -o source/cache/CMakeFiles/cache-objects.dir/memcache.cpp.o -c /tmp/pinches/EasyBuild/UDA/2.7.5/intel-compilers-2023.2.1/UDA-release-2.7.5/source/cache/memcache.cpp
error: unknown warning option '-Wno-use-after-free' [-Werror,-Wunknown-warning-option]

[jholloc]

I've found https://estuarine.jp/2021/03/install-oneapi which gives details of installing the Intel compilers on Ubuntu. I'll try this on my local VM and if it works i'll set up an Intel CI build.

[jholloc]

@SimonPinches i've added an Intel CI build and made the required fixes to get the Intel build working.

[SimonPinches]

Thanks. I'll give it a try as soon as I get a chance.

[SimonPinches]

Thanks @jholloc. I confirm this now works for our intel/2023b toolchains on RHEL9 and so will approve this PR again :-)

[stephen-dixon]

UKAEA client and server builds and tests pass