Bug in poisson2_example

[mmoelle1]

Dear all,

there seems to be a bug in the poisson2_example.cpp file which only shows when compiled in Debug mode. The code compiles fine but give the following error when run:

Loaded file /DO_NOT_SHOW_HOME_DIR_PATH/gismo.stable/filedata/pde/poisson2d_bvp.xml
Source function [ 2*pi^2*sin(pi*x)*sin(pi*y) ]
Boundary conditions:
* Dirichlet     : 6

Patches: 2, degree: 2
(dot1=assembled, dot2=solved, dot3=got_error)

DoFs: 10libc++abi.dylib: terminating with uncaught exception of type std::runtime_error: GISMO_ENSURE `m_curElimId==0` finalize() was not called on gsDofMapper
/DO_NOT_SHOW_HOME_DIR_PATH/gismo.stable/src/gsCore/gsDofMapper.h, line 408 (boundarySize)
Abort trap: 6

Any help is appreciated.

Matthias

[filiatra]

Hi,
I cannot reproduce this.. on linux in Debug mode it seems to run fine..
Even valgrind finds no flaws...

[mmoelle1]

Hi, This error occurred under macOS with the Apple Clang compiler. I am now double-checking with GCC (same OS, same configuration) to check if it is only a problem with Clang.

…

On 26 Jun 2018, at 11:53, Angelos Mantzaflaris ***@***.***> wrote: Hi, I cannot reproduce this.. on linux in Debug mode it seems to run fine.. Even valgrind finds no flaws... — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#200 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB5nozglZToXGTReunMoiYaeEEV-5tqcks5uAgSWgaJpZM4U2YC3>.

[filiatra]

Ok also trying clang on linux right now

[mmoelle1]

Here is a small update to the problem. It also happens with GCC (6.4.0) under macOS. I could narrow down the error to the following: The push()-method of the file gsExprAssembler.h is called twice. During the first call, colMap.isFinalized() and rowMap.isFinalized() both return true. The push()-method is called a second time and then, colMap.isFinalized() returns false while rowMap.isFinalized() returns true. That exception is thrown at this line GISMO_ASSERT(colMap.boundarySize() == fixedDofs.rows(), …); Matthias

…

On 26 Jun 2018, at 13:10, Angelos Mantzaflaris ***@***.***> wrote: Ok also trying clang on linux right now — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#200 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB5no5RLHMELX3-USCsu-7wdAWv7sFgoks5uAha0gaJpZM4U2YC3>.

[mmoelle1]

I forgot to mention that the value of m_curElimID in method boundarySize() is this: boundarySize m_curElimId: 1883329032 Seems as if things remain uninitialised.

…

On 26 Jun 2018, at 13:27, Matthias Möller ***@***.***> wrote: Here is a small update to the problem. It also happens with GCC (6.4.0) under macOS. I could narrow down the error to the following: The push()-method of the file gsExprAssembler.h is called twice. During the first call, colMap.isFinalized() and rowMap.isFinalized() both return true. The push()-method is called a second time and then, colMap.isFinalized() returns false while rowMap.isFinalized() returns true. That exception is thrown at this line GISMO_ASSERT(colMap.boundarySize() == fixedDofs.rows(), …); Matthias > On 26 Jun 2018, at 13:10, Angelos Mantzaflaris ***@***.*** ***@***.***>> wrote: > > Ok also trying clang on linux right now > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub <#200 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB5no5RLHMELX3-USCsu-7wdAWv7sFgoks5uAha0gaJpZM4U2YC3>. >

[hverhelst]

@mmoelle1 @filiatra can we close this?