[ublas] Mass test failures on msvc12 and other on all msvc versions
Hi, Nearly all tests fails for msvc12 because of the use of constexpr keyword. Could this be fixed? I prepared a pull request nearly a month ago: https://github.com/boostorg/ublas/pull/5 Btw, in those tests failing for all MSVC versions an assertion failure is reported by the STL: map/set iterators incompatible. Assertion failures should be visible in the regression results, shouldn't they? At least in the file test_inplace_solve.cpp (of course with -DUSE_MAPPED_VECTOR_OF_MAPPED_VECTOR) this assertion failure is caused by the matrix_sparse.hpp line 2002: return it_ == it.it_; Which is called by: triangular.hpp line 1328 matrix_expression.hpp line 1598 iterator.hpp line 185 ... test_inplace_solve.cpp line 72 I hope that helps. Do you plan to release the version of uBlas from develop in 1.56? Regards, Adam
On Mon, Jun 9, 2014 at 7:47 AM, Adam Wulkiewicz
Hi,
Nearly all tests fails for msvc12 because of the use of constexpr keyword. Could this be fixed? I prepared a pull request nearly a month ago: https://github.com/boostorg/ ublas/pull/5
Adam, when there is no response to a pull request, please post a message to the Community Maintenance Team mailing list so they can look into it. Be sure to include the github link. Thanks for the regression test failure reports you have been posting! --Beman
On Mon, Jun 9, 2014 at 7:47 AM, Adam Wulkiewicz
wrote: Hi,
Nearly all tests fails for msvc12 because of the use of constexpr keyword. Could this be fixed? I prepared a pull request nearly a month ago: https://github.com/boostorg/ ublas/pull/5
Adam, when there is no response to a pull request, please post a message to the Community Maintenance Team mailing list so they can look into it. Be sure to include the github link. There was response however the pull request wasn't merged nor other
Beman Dawes wrote: patch applied since then. In that case, should I also inform CMT? Regards, Adam
whoops ! Yes, indeed, we're a little bit behind to say the least. I'll try
to merge that today or tomorrow at maximum.
No we don't plan to release anything for 1.56 as we have a few projects
cooking but not yet ready due to ... well, ... you know job, kids, etc...
;-)
Sorry guys about that...
On Mon, Jun 9, 2014 at 12:47 PM, Adam Wulkiewicz
Hi,
Nearly all tests fails for msvc12 because of the use of constexpr keyword. Could this be fixed? I prepared a pull request nearly a month ago: https://github.com/boostorg/ ublas/pull/5
Btw, in those tests failing for all MSVC versions an assertion failure is reported by the STL:
map/set iterators incompatible.
Assertion failures should be visible in the regression results, shouldn't they?
At least in the file test_inplace_solve.cpp (of course with -DUSE_MAPPED_VECTOR_OF_MAPPED_VECTOR) this assertion failure is caused by the
matrix_sparse.hpp line 2002:
return it_ == it.it_;
Which is called by: triangular.hpp line 1328 matrix_expression.hpp line 1598 iterator.hpp line 185 ... test_inplace_solve.cpp line 72
I hope that helps.
Do you plan to release the version of uBlas from develop in 1.56?
Regards, Adam
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/ mailman/listinfo.cgi/boost
[Adam Wulkiewicz]
Btw, in those tests failing for all MSVC versions an assertion failure is reported by the STL: map/set iterators incompatible. Assertion failures should be visible in the regression results, shouldn't they? At least in the file test_inplace_solve.cpp (of course with -DUSE_MAPPED_VECTOR_OF_MAPPED_VECTOR) this assertion failure is caused by the matrix_sparse.hpp line 2002: return it_ == it.it_;
This assertion is complaining that C++98-11 forbids iterator comparisons between unparented or differently-parented iterators. C++14 and VS14 CTP1 permit value-initialized forward-or-better iterators to be compared to each other (and they are equal), but they still can't be compared to iterators with parents. STL
participants (4)
-
Adam Wulkiewicz
-
Beman Dawes
-
David Bellot
-
Stephan T. Lavavej