Noticeable fixed issues that may affect the results obtained with previous versions

Ticket #256 reported that the scalar product of two unsymmetric tensors was not properly computed.

This may affect single crystal finite strain computations to a limited extent, as the Mandel stress tensor is almost symmetric.

Tickets fixed

Ticket #256: Scalar products of unsymmetric tensors

The scalar product of unsymmetric tensors was not computed correctly in \(2D\) and \(3D\).

For more details, see: https://sourceforge.net/p/tfel/tickets/256/

Ticket #254: Register libraries for ldconfig

In previous versions, the build version and the API version of shared libraries were not specified.

For more details, see: https://sourceforge.net/p/tfel/tickets/254/

Ticket #246: Explicit implementation of the Berveiller-Zaoui homogeneisation scheme is inefficient

Several polycrystal behaviours based on the Berveiller-Zaoui homogeneisation scheme are available in the tests of MFront.

For some unknown reasons, those implementations declared the equivalent plastic strain as an auxiliary state variable. This lead to a significantly more difficult convergence of the Runge-Kutta algorithm.

Since the results are correct, those implementations are kept as they are, but new corrected implementations are provided. Appropriate warnings are added to the description of flawed behaviours. Users are advised to use the newly proposed implementation.

For more details, see: https://sourceforge.net/p/tfel/tickets/246/

The following headers were not installed by cmake:

For more details, see: https://sourceforge.net/p/tfel/tickets/244/

Ticket #239: Missing ostream header for gcc-10.1.0

For more details, see: https://sourceforge.net/p/tfel/tickets/239/

Ticket #238: getPartialJacobianInvert is broken for arrays of internal state variables

For more details, see: https://sourceforge.net/p/tfel/tickets/238/

Ticket #224: Fix compilation of the gentype3.cxx test in C++-17 with gcc-9.2

For more details, see: https://sourceforge.net/p/tfel/tickets/208/

Ticket #208: Support of the @GasEquationOfState keyword is broken in MTest

MTest is able to model a tight pipe. This hypothesis assumes that the amount of substance of the gas inside the pipe is constant. The evolution of the inner pressure is linked to the temperature and radius evolution through the gas equation of state. By default, the equation of state of a perfect gas is used.

The @GasEquationOfState keyword was introduced to specify the equation of state of the gas. This feature has never been tested and was not functional in previous versions.

For more details, see: https://sourceforge.net/p/tfel/tickets/208/

Ticket #207: Circumvent a possible bug in FindBoost for cmake-3.16

Two boost components are searched when enabling python bindings: boost-python and boost-numpy. The first one is required and the second one is optional.

The detection process that was used in previous versions called find_package for two components at the same time and then we checkd of two components were found.

In cmake 3.16 (at least), the behaviour of find_package changed and if one component is missing, both are declared not found.

We now check for boost-python first and then search for boost-numpy.

For more details, see: https://sourceforge.net/p/tfel/tickets/207/

Ticket #197: Allow using flags defined in standard environment variables to build TFEL/MFront

The variable USE_EXTERNAL_COMPILER_FLAGS can now be defined in cmake to use the flags defined in the environment variables CFLAGS and CXXFLAGS.

For more details, see: https://sourceforge.net/p/tfel/tickets/197/