Core Modules issueshttps://gitlab.dune-project.org/groups/core/-/issues2017-08-29T14:57:11Zhttps://gitlab.dune-project.org/core/dune-geometry/-/issues/12Update composite quadraturerule to use the improved refinement interface2017-08-29T14:57:11ZJö Fahlkejorrit@jorrit.deUpdate composite quadraturerule to use the improved refinement interfacePlease update `quadraturerules/compositequadraturerule.hh` to use the new refinement interface. It you feel like it, you may also update it to also provide the new refinement interface.Please update `quadraturerules/compositequadraturerule.hh` to use the new refinement interface. It you feel like it, you may also update it to also provide the new refinement interface.https://gitlab.dune-project.org/core/dune-localfunctions/-/issues/12Consecutive failures may lead to a "success" in unit tests2018-12-01T20:39:19ZSantiago Ospina De Los Ríossospinar@gmail.comConsecutive failures may lead to a "success" in unit testsI have noticed you use the following structure when checking the success of a unit tests:
```c++
bool success = true;
success &= some_complicated_test(); // return success
success &= another_complicated_test(); // return success
.....I have noticed you use the following structure when checking the success of a unit tests:
```c++
bool success = true;
success &= some_complicated_test(); // return success
success &= another_complicated_test(); // return success
...
return !success;
```
This is problematic because certain patterns of failures will lead `success` to be `true` (e.g. pair number of failures). It may be corrected with the following structure.
```c++
bool failure = false;
failure |= some_complicated_test(); // return failure
failure |= another_complicated_test(); // return failure
...
return failure;
```https://gitlab.dune-project.org/core/dune-istl/-/issues/10Matrix Iterator fails to build with clang 3.52017-05-21T00:30:58ZTimo KochMatrix Iterator fails to build with clang 3.5Fails with clang 3.5 throwing the following error (gcc does not complain):
```
/temp/timok/buildbot/balu/clang35-dune-master/dune-istl/dune/istl/matrix.hh:409:9: error: use of non-static data member 'columns_' of 'DenseMatrixBase' fr...Fails with clang 3.5 throwing the following error (gcc does not complain):
```
/temp/timok/buildbot/balu/clang35-dune-master/dune-istl/dune/istl/matrix.hh:409:9: error: use of non-static data member 'columns_' of 'DenseMatrixBase' from nested type 'ConstIterator'
columns_ = 0;
^~~~~~~~
```
Changing to `Iterator::columns_ = 0;` fixes it for me. Is that the correct fix?https://gitlab.dune-project.org/core/dune-grid/-/issues/36CheckSubEntityGeometry forces grid implementors to sublass Entity2017-04-07T14:32:03ZMarkus BlattCheckSubEntityGeometry forces grid implementors to sublass EntityIs that on purpose?
Previously it was possible for grids to use their own implementation without inheriting from Dune::Entity. That seemed a lot more convenient.Is that on purpose?
Previously it was possible for grids to use their own implementation without inheriting from Dune::Entity. That seemed a lot more convenient.https://gitlab.dune-project.org/core/dune-common/-/issues/57test failures on 32bit architectures2017-08-04T09:59:31ZAnsgar Burchardtansgar.burchardt@tu-dresden.detest failures on 32bit architecturesFrom Debian's [2.5.0rc1 build logs] it looks like `sllisttest` fails on all 32bit architectures (though not all builds are finished yet).
`poolallocatortest` fails on at least i386.
[2.5.0rc1 build logs]: https://buildd.debian.org/st...From Debian's [2.5.0rc1 build logs] it looks like `sllisttest` fails on all 32bit architectures (though not all builds are finished yet).
`poolallocatortest` fails on at least i386.
[2.5.0rc1 build logs]: https://buildd.debian.org/status/logs.php?pkg=dune-common&ver=2.5.0~rc1-1DUNE 2.5.0https://gitlab.dune-project.org/core/dune-geometry/-/issues/13Introduction of value semantics for ReferenceElement changed semantics of int...2017-09-19T17:10:56ZCarsten Gräsergraeser@math.fau.deIntroduction of value semantics for ReferenceElement changed semantics of integrationOuterNormal()Before the restructuring the methods `integrationOuterNormal()`, `position()` and `type()` of `ReferenceElement` returned const references to cached values. After the change these methods all return copies. While it seems natural to retu...Before the restructuring the methods `integrationOuterNormal()`, `position()` and `type()` of `ReferenceElement` returned const references to cached values. After the change these methods all return copies. While it seems natural to return values if the reference elements are returned by values themselves, it breaks compatibility.
A consequence of this are the broken tests in dune-localfunctions. There, the Raviart-Thomas elements store pointers to the result of `integrationOuterNormal()`. While this could be adjusted in dune-localfunctions, users may also run into the same problem breaking their code without an error message.https://gitlab.dune-project.org/core/dune-localfunctions/-/issues/13CI fails with clang >= 62019-01-10T11:04:22ZCarsten Gräsergraeser@math.fau.deCI fails with clang >= 6Since centralising the CI in 398088b339bc4 it fails for newer clang versions (clang-6.0 on Ubuntu 18.04 and clang-7 on Debian 10):
https://gitlab.dune-project.org/core/dune-localfunctions/pipelines/15022
I can reproduce the problem ...Since centralising the CI in 398088b339bc4 it fails for newer clang versions (clang-6.0 on Ubuntu 18.04 and clang-7 on Debian 10):
https://gitlab.dune-project.org/core/dune-localfunctions/pipelines/15022
I can reproduce the problem locally (Ubuntu 18.04, same compiler) using the same compile flags as in the CI setup. `-O3` triggers the problem whily `-O0` does not.
The problem is a floating point exception which seems to be in a call to `ddp` within `lagrange/qk/qklocalbasis.hh` triggered by the check of `QkLocalFiniteElement<double,double,1,1>` in
`lagrangeshapefunctiontest.cc`. Looking at the code I cannot see why there should be a floating point problem as all branches in the function look non-problematic to me. Unfortunately debugging this is hard since many values are optimized out. What's IMHO strange is, that I cannot run the generated file in valgrid due to an `vex amd64->IR: unhandled instruction bytes: ...` although the compile flags do not enable any special target architecture/instruction set.https://gitlab.dune-project.org/core/dune-istl/-/issues/9Build error in dune-istl2017-05-21T00:30:58ZGovind SahaiBuild error in dune-istlHi to all. I am new here. I am getting a build error while doing `./dune-common/bin/dunecontrol all` in file `/dune-istl/dune/istl/tutorial/example.cc`.
```
/home/mnciitbhu/Git_Clones/__/dune/dune-istl/dune/istl/tutorial/example.cc: I...Hi to all. I am new here. I am getting a build error while doing `./dune-common/bin/dunecontrol all` in file `/dune-istl/dune/istl/tutorial/example.cc`.
```
/home/mnciitbhu/Git_Clones/__/dune/dune-istl/dune/istl/tutorial/example.cc: In function ‘void test_IO()’:
/home/mnciitbhu/Git_Clones/__/dune/dune-istl/dune/istl/tutorial/example.cc:301:13: error: ‘using Iterator = class Dune::VariableBlockVector<Dune::FieldVector<double, 2> >::RealIterator<Dune::BlockVector<Dune::FieldVector<double, 2> >, Dune::BlockVectorWindow<Dune::FieldVector<double, 2>, std::allocator<Dune::FieldVector<double, 2> > >&> {aka class Dune::VariableBlockVector<Dune::FieldVector<double, 2> >::RealIterator<Dune::BlockVector<Dune::FieldVector<double, 2> >, Dune::BlockVectorWindow<Dune::FieldVector<double, 2>, std::allocator<Dune::FieldVector<double, 2> > >&>}’ has no member named ‘index’
*i = (i.index()%3)+1;
^
```
Am I doing something wrong here? I am using `Fedora 23` and `gcc 5.3.1`https://gitlab.dune-project.org/core/dune-grid/-/issues/35headercheck does not compile anymore.2017-04-07T14:32:03ZRobert Kheadercheck does not compile anymore.There have been some recent changes in dune/grid/test and dune/grid/identitygrid that make the headercheck fail.
@linus.seelinger : Please fix the test/yasp things
@oliver.sander : Please fix the identitygrid things
or use cmake macro...There have been some recent changes in dune/grid/test and dune/grid/identitygrid that make the headercheck fail.
@linus.seelinger : Please fix the test/yasp things
@oliver.sander : Please fix the identitygrid things
or use cmake macro
exclude_from_headercheck( header-name )
to remove failing headers from the list of headers to be tested.DUNE 2.5.0https://gitlab.dune-project.org/core/dune-common/-/issues/55vectorization interface causes Alberta regression2017-05-25T21:04:48ZJö Fahlkejorrit@jorrit.devectorization interface causes Alberta regressionOriginally posted by @felix.gruber in !121:
@joe: this merge broke dune-grid with Alberta. Somehow the new `swap(T& v1, T& v2, bool mask)` prevents `std::swap(T& v1, T& v2)` to be found.
The error message I get with g++-5.2.0 is
```
/h...Originally posted by @felix.gruber in !121:
@joe: this merge broke dune-grid with Alberta. Somehow the new `swap(T& v1, T& v2, bool mask)` prevents `std::swap(T& v1, T& v2)` to be found.
The error message I get with g++-5.2.0 is
```
/home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/elementinfo.hh:391:11: error: no matching function for call to ‘swap(Dune::Alberta::ElementInfo<1>::Instance*&, Dune::Alberta::ElementInfo<1>::Instance*&)’
swap( instance_, other.instance_ );
^
In file included from /home/data/gruber/src/dune-git/dune-common/dune/common/densematrix.hh:19:0,
from /home/data/gruber/src/dune-git/dune-common/dune/common/fmatrix.hh:15,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/agrid.hh:25,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid.hh:5,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/dgfparser.hh:8,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/dgfparser.cc:10:
/home/data/gruber/src/dune-git/dune-common/dune/common/simd.hh:300:8: note: candidate: template<class T> void Dune::swap(T&, T&, bool)
void swap(T &v1, T &v2, bool mask)
^
/home/data/gruber/src/dune-git/dune-common/dune/common/simd.hh:300:8: note: template argument deduction/substitution failed:
In file included from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/meshpointer.hh:15:0,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/coordcache.hh:6,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/agrid.hh:42,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid.hh:5,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/dgfparser.hh:8,
from /home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/dgfparser.cc:10:
/home/data/gruber/src/dune-git/dune-grid/dune/grid/albertagrid/elementinfo.hh:391:11: note: candidate expects 3 arguments, 2 provided
swap( instance_, other.instance_ );
^
```
and looking at the code in `dune/grid/albertagrid/elementinfo.hh:391`
```c++
template< int dim >
inline ElementInfo< dim >::ElementInfo ( ElementInfo &&other )
: instance_( NULL )
{
using namespace std;
swap( instance_, other.instance_ );
}
```
I do not understand why `std::swap` isn't used, especially since `<utility>` gets included at the top of the file so that it's definition should be available for overload resolution.Jö Fahlkejorrit@jorrit.deJö Fahlkejorrit@jorrit.dehttps://gitlab.dune-project.org/core/dune-geometry/-/issues/14Redesign of reference elements not documented in recent changes2017-11-29T17:56:56ZCarsten Gräsergraeser@math.fau.deRedesign of reference elements not documented in recent changesThe changes done in !52 should be mentioned in the 'recent changes'/'release notes'. We should especially highlight the non-compatible changes.The changes done in !52 should be mentioned in the 'recent changes'/'release notes'. We should especially highlight the non-compatible changes.DUNE 2.6.0Steffen Müthingsteffen.muething@iwr.uni-heidelberg.deSteffen Müthingsteffen.muething@iwr.uni-heidelberg.de2017-11-10https://gitlab.dune-project.org/core/dune-localfunctions/-/issues/14Generic finite element implementations rely on undefined behaviour2022-09-23T13:11:09ZCarsten Gräsergraeser@math.fau.deGeneric finite element implementations rely on undefined behaviourAll 'generic' finite element implementations, i.e., those building up on the infrastructure in `dune/localfunctions/utility/` make heavy use of `reinterpret_cast` breaking strict aliasing rules.
As a consequence evaluation of the basis f...All 'generic' finite element implementations, i.e., those building up on the infrastructure in `dune/localfunctions/utility/` make heavy use of `reinterpret_cast` breaking strict aliasing rules.
As a consequence evaluation of the basis functions leads to undefined behaviour.
For those not aware of strict aliasing errors: A typical manifestation of this is, that optimized code may silently compute wrong numbers in special cases because the compiler makes wrong assumptions on when to sync register values with memory.https://gitlab.dune-project.org/core/dune-istl/-/issues/8dune-istl 2.4.1 Coordination Task2017-05-21T00:30:58ZSteffen Müthingsteffen.muething@iwr.uni-heidelberg.dedune-istl 2.4.1 Coordination TaskI'll use this task to track the backports for dune-common 2.4.1.
If you have suggestions for more fixes or concerns about one of the backports, please comment.
## Backported commits
- [x] f40f330 [bugfix][RestartedGMResSolver] s...I'll use this task to track the backports for dune-common 2.4.1.
If you have suggestions for more fixes or concerns about one of the backports, please comment.
## Backported commits
- [x] f40f330 [bugfix][RestartedGMResSolver] signature of method apply was wrong when real_type != double.
- [x] 8fe92a2 Git ignore build-* instead of build-cmake.
- [x] 61af51d [cleanup] Remove leftover (?) file Doxydep
- [x] b1882eb [galerkintest] Fix output of matrix size.
- [x] 9fe002f [bugfix] Use correct integral type in the graph classes.
- [x] cf48329 [release][repartition][bugfix] correct implementation of usage of ParMetis::idx_t for different versions of ParMetis.
- [x] d1da71f Make sure the vectors x and b handed to the 'apply' method have the correct size
- [x] 6de3455 [tests] Catch Dune exceptions to ease identifying problems.
- [x] e25671c [superlu] Ensure vectors x and b handed to 'apply' have same size
- [x] e95bfb9 [tests] Use correct sized rhs for Superlu/UMFPack sub-matrix tests
- [x] 6c9aa54 Merge branch 'feature/FS1659-Matlab-output-of-DynamicMatrix' into 'master'
- [x] 6c51c8f [bugfix] Choose a gcc4.4 compliant header to test for dune-istl
- [x] 8b2910b [test] Adjust exception handling in tests.
- [x] 26a591e Fix headercheck
- [x] e85759d Merge branch 'feature/smaller-cleanups' into 'master'
- [x] 2fdafaa Remove an out-commented test for a way to build a BCRSMatrix that won't work anyway
- [x] 922bd81 Merge branch 'feature/fix-shadow-warnings' into 'master'
- [x] 8d26a70 [CMake] Do not require BLAS
- [x] 9bf96e0 Merge branch 'feature/kamg-reference' into 'master'
- [x] a8f68be Provide operator[] for integral_constant<std::size_t, k> (needs some porting work)
- [x] 0196c61 Provide operator[] with integral_constant<std::size_t, k> for MultiTypeBlockMatrix (needs some porting work)
- [x] 3aed940 [bugfix] fix broken assertion.
- [x] 1b544df Merge branch 'matrix-loadbalance-to-more' into 'master'
## Merge requests and bugs
- [x] !25 Backport operator[] for MultiTypeBlockVector and MultiTypeBlockMatrix
- [x] !28 Support load balance of global matri from one to more pocesses.DUNE 2.4.1Steffen Müthingsteffen.muething@iwr.uni-heidelberg.deSteffen Müthingsteffen.muething@iwr.uni-heidelberg.dehttps://gitlab.dune-project.org/core/dune-grid/-/issues/34cmake doesn't inform whether UG is seq. or parallel2017-04-07T14:32:03ZChristian Engwercmake doesn't inform whether UG is seq. or parallelIn former times configure told us, whether UG was compiled with or without MPI. cmake doesn't tell this anymore, which is a bit annoying.In former times configure told us, whether UG was compiled with or without MPI. cmake doesn't tell this anymore, which is a bit annoying.DUNE 2.5.0https://gitlab.dune-project.org/core/dune-common/-/issues/52Update man page of dunecontrol2017-08-04T09:59:31ZOliver Sanderoliver.sander@tu-dresden.deUpdate man page of dunecontrolIt still mentions autogen and the AutoTools in a few places.It still mentions autogen and the AutoTools in a few places.DUNE 2.5.0https://gitlab.dune-project.org/core/dune-geometry/-/issues/15clang fails to compile topologyfactory.hh2017-11-03T12:19:43ZJanick Gerstenbergerclang fails to compile topologyfactory.hhclang (ubunutu:16.4,debian:9) fails with the following errors
```sh
/duneci/modules/dune-geometry/dune/geometry/topologyfactory.hh:133:15: error: member reference base type 'Object *' (aka 'const typename Factory::Object *') is not a st...clang (ubunutu:16.4,debian:9) fails with the following errors
```sh
/duneci/modules/dune-geometry/dune/geometry/topologyfactory.hh:133:15: error: member reference base type 'Object *' (aka 'const typename Factory::Object *') is not a structure or union
object.reset( Factory::create( gt, key ) );
~~~~~~^~~~~~
/duneci/modules/dune-geometry/dune/geometry/topologyfactory.hh:134:20: error: member reference base type 'Object *' (aka 'const typename Factory::Object *') is not a structure or union
return object.get();
~~~~~~^~~~
```https://gitlab.dune-project.org/core/dune-localfunctions/-/issues/15RT1***LocalFiniteElement returns wrong values2020-08-31T14:14:25ZOliver Sanderoliver.sander@tu-dresden.deRT1***LocalFiniteElement returns wrong valuesUnit-testing the classes `RT1Cube2DLocalFiniteElement` and `RT12DLocalFiniteElement` uncovered run-time / wrong-result bugs in both of them: (MR !127)
Bug in LocalInterpolation for finite element type Dune::RT1Cube2DLocalFiniteEleme...Unit-testing the classes `RT1Cube2DLocalFiniteElement` and `RT12DLocalFiniteElement` uncovered run-time / wrong-result bugs in both of them: (MR !127)
Bug in LocalInterpolation for finite element type Dune::RT1Cube2DLocalFiniteElement<double, double> const
Degree of freedom 0 applied to shape function 0 yields value 0, not the expected value 1
Bug in LocalInterpolation for finite element type Dune::LocalFiniteElementVirtualImp<Dune::RT1Cube2DLocalFiniteElement<double, double> > const
Degree of freedom 0 applied to shape function 0 yields value 0, not the expected value 1
testFE(rt1cube2dlfemDedicated) failed
Bug in LocalInterpolation for finite element type Dune::RT12DLocalFiniteElement<double, double> const
Degree of freedom 0 applied to shape function 0 yields value 0, not the expected value 1
Bug in LocalInterpolation for finite element type Dune::LocalFiniteElementVirtualImp<Dune::RT12DLocalFiniteElement<double, double> > const
Degree of freedom 0 applied to shape function 0 yields value 0, not the expected value 1
testFE(rt12dlfemDedicated) failed
That's quite dangerous: You may not even notice that your RT problems are assembled incorrectly.
@gruenich , didn't you write those? Could you have a look?https://gitlab.dune-project.org/core/dune-istl/-/issues/6umfpacktest.cc does not compile if umfpack is present but not suitesparse2017-05-21T00:30:58ZCarsten Gräsergraeser@math.fau.deumfpacktest.cc does not compile if umfpack is present but not suitesparseAs far as I understand the test the problem is the following:
In this case the test defined SuiteSparse_UMFPACK_FOUND which activates the test.
On the other hand HAVE_SUITESPARSE_UMFPACK which is checked in our umfpack.hh is set to...As far as I understand the test the problem is the following:
In this case the test defined SuiteSparse_UMFPACK_FOUND which activates the test.
On the other hand HAVE_SUITESPARSE_UMFPACK which is checked in our umfpack.hh is set to ENABLE_SUITESPARSE in config.h. This, however, is only set if SuiteSparse_FOUND is true.
DUNE 3.0.0Christoph GrüningerChristoph Grüningerhttps://gitlab.dune-project.org/core/dune-grid/-/issues/33test-parallel-ug-mpi-2 fails with clang2017-04-07T14:32:03ZCarsten Gräsergraeser@math.fau.detest-parallel-ug-mpi-2 fails with clangWhen compiled with clang-3.8 and libstdc++-6.2 (Ubuntu 16.10) the test `test-parallel-ug` fails for 2 mpi ranks with the error
```
--------------------------------------------------------------------------
mpirun noticed that proces...When compiled with clang-3.8 and libstdc++-6.2 (Ubuntu 16.10) the test `test-parallel-ug` fails for 2 mpi ranks with the error
```
--------------------------------------------------------------------------
mpirun noticed that process rank 1 with PID 12585 on node patton exited on signal 4 (Illegal instruction).
--------------------------------------------------------------------------
```
Notice that this is not for master but for !101 because UG does not work with clang otherwise.https://gitlab.dune-project.org/core/dune-common/-/issues/51dunecontrol: improper quoting of command arguments2017-05-25T21:04:49ZJö Fahlkejorrit@jorrit.dedunecontrol: improper quoting of command argumentsThis means you are limited to very simple skripts in `dunecontrol exec`/`dunecontrol bexec` that don't use quotes themselves.
```
joe@paranoia:~/Projekte/dune-quote/dune-common$ LC_ALL=C bin/dunecontrol exec 'var="quoted value"; echo...This means you are limited to very simple skripts in `dunecontrol exec`/`dunecontrol bexec` that don't use quotes themselves.
```
joe@paranoia:~/Projekte/dune-quote/dune-common$ LC_ALL=C bin/dunecontrol exec 'var="quoted value"; echo "$var"'
--- going to build dune-common ---
--- calling exec for dune-common ---
bash: value: command not found
--- dune-common done ---
--- done ---
```