GCC 13 errors
sagitter opened this issue · 7 comments
sagitter commented
Does this problem persist on the current master?
- I have verified the issue on the current master
Is there an existing issue for this?
- I have searched the existing issues
Current Behavior
In file included from /builddir/build/BUILD/seqan3-c5ec15bd0682eae350892e120773cf639904f71d/include/seqan3/alignment/matrix/detail/trace_iterator_base.hpp:19,
from /builddir/build/BUILD/seqan3-c5ec15bd0682eae350892e120773cf639904f71d/include/seqan3/alignment/matrix/detail/trace_iterator.hpp:15,
from /builddir/build/BUILD/seqan3-c5ec15bd0682eae350892e120773cf639904f71d/test/unit/alignment/matrix/detail/aligned_sequence_builder_test.cpp:16:
/builddir/build/BUILD/seqan3-c5ec15bd0682eae350892e120773cf639904f71d/include/seqan3/alignment/matrix/detail/two_dimensional_matrix_iterator_base.hpp: In instantiation of 'class seqan3::detail::two_dimensional_matrix_iterator_base<seqan3::detail::two_dimensional_matrix<seqan3::detail::trace_directions>::basic_iterator<false>, seqan3::detail::matrix_major_order::row>':
/builddir/build/BUILD/seqan3-c5ec15bd0682eae350892e120773cf639904f71d/include/seqan3/alignment/matrix/detail/two_dimensional_matrix.hpp:324:60: required from 'class seqan3::detail::two_dimensional_matrix<seqan3::detail::trace_directions>::basic_iterator<false>'
/builddir/build/BUILD/seqan3-c5ec15bd0682eae350892e120773cf639904f71d/test/unit/alignment/matrix/detail/aligned_sequence_builder_test.cpp:61:85: required from here
/builddir/build/BUILD/seqan3-c5ec15bd0682eae350892e120773cf639904f71d/include/seqan3/alignment/matrix/detail/two_dimensional_matrix_iterator_base.hpp:85:18: error: redeclaration of 'template<class derived_t, seqan3::detail::matrix_major_order order> class seqan3::detail::two_dimensional_matrix_iterator_base' with different constraints
85 | friend class two_dimensional_matrix_iterator_base;
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Expected Behavior
No errors
Steps To Reproduce
Compiling seqan3 c5ec15b in Fedora 39's RPM build system
Environment
- Operating system: Fedora 39 (devel branch) any architecture
- SeqAn version: c5ec15bd0682eae350892e120773cf639904f71d
- Compiler: GCC-13.0.1Anything else?
Full log attached
seqan3.zip
eseiler commented
Hey @sagitter,
Whats the gcc commit?
We have nightly builds with the gcc master, and the error doesn't show up
EDIT: Was able to reproduce
sagitter commented
Hey @sagitter,
Hi Enrico.
Sorry for the late reply
Whats the gcc commit? We have nightly builds with the gcc master, and the error doesn't show up
I don't know but you can find a Changelog of the GCC RPM (used in this seqan build) here that reports the ID of fixed bug
EDIT: Was able to reproduce
I will compile seqan newly in some days.
Thank you
eseiler commented
There are some new errors, I'll leave this issue open until the errors are fixed.
sagitter commented
From commit 174255e
/builddir/build/BUILD/seqan3-174255eea53d4847e7ce8cb0dc29415f177dc54d/test/unit/alignment/pairwise/pairwise_alignment_collection_test_template.hpp:45:48: required from 'void gtest_suite_pairwise_alignment_collection_test_::score<gtest_TypeParam_>::TestBody() [with gtest_TypeParam_ = pairwise_alignment_fixture<(& seqan3::test::alignment::collection::simd::global::affine::banded::dna4_all_same)>]'
/builddir/build/BUILD/seqan3-174255eea53d4847e7ce8cb0dc29415f177dc54d/test/unit/alignment/pairwise/pairwise_alignment_collection_test_template.hpp:40:1: required from here
/builddir/build/BUILD/seqan3-174255eea53d4847e7ce8cb0dc29415f177dc54d/include/seqan3/alignment/pairwise/detail/pairwise_alignment_algorithm_banded.hpp:368:9: error: possibly dangling reference to a temporary [-Werror=dangling-reference]
368 | for (auto && alphabet2 : sequence2 | std::views::drop(1))
| ^~~
/builddir/build/BUILD/seqan3-174255eea53d4847e7ce8cb0dc29415f177dc54d/include/seqan3/alignment/pairwise/detail/pairwise_alignment_algorithm_banded.hpp:368:44: note: the temporary was destroyed at the end of the full expression 'std::ranges::views::__adaptor::operator|<_Partial<std::ranges::views::_Drop, int>, std::span<__vector(1) int, 18446744073709551615>&>((* & sequence2), ((const std::ranges::views::__adaptor::_RangeAdaptor<std::ranges::views::_Drop>*)(& std::ranges::views::drop))->std::ranges::views::__adaptor::_RangeAdaptor<std::ranges::views::_Drop>::operator()<int>(1))'
368 | for (auto && alphabet2 : sequence2 | std::views::drop(1))
| ~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~
cc1plus: all warnings being treated as errors
gmake[2]: Leaving directory '/builddir/build/BUILD/seqan3-174255eea53d4847e7ce8cb0dc29415f177dc54d/build'
gmake[2]: *** [alignment/pairwise/CMakeFiles/global_affine_banded_collection_simd_test.dir/build.make:79: alignment/pairwise/CMakeFiles/global_affine_banded_collection_simd_test.dir/global_affine_banded_collection_simd_test.cpp.o] Error 1
Full log:
seqan-git174255ee.zip
With GCC:
* Fri Mar 24 2023 gcc-13.0.1-0.10
- update from trunk
- PRs analyzer/109239, c++/105996, c++/108390, c++/108795, c++/109177,
c++/109241, c/84900, fortran/104572, middle-end/109258,
modula2/107630, modula2/109248, modula2/109264, target/103628,
target/109137, target/109228, target/109244, testsuite/105959,
tree-optimization/99739, tree-optimization/107569,
tree-optimization/109176, tree-optimization/109262
eseiler commented
Yep, there is also a (bogus? ) value-may-be-used-uninitialized and an ICE there. Will probably take a while to fix.
eseiler commented
Yes, the ICE I saw is something weird.
It seems to be related to GoogleTest, SIMD, and LTO.
It only affects:
global_affine_banded_collection_simd_test.cpp
global_affine_unbanded_collection_simd_test.cpp
global_affine_unbanded_collection_simd_aa27_test.cpp
If I recall correctly, this happens during linking, not compiling.
For now, I just removed the LTO flags (-flto, -ffat-lto-objects).
Anyway, this issue is fixed :)