Normann Rion authored Jun 15, 2021 and Sébastien Villemot committed Jun 15, 2021

Partially addresses issue #1680:
- unconditional welfare resorts to dynare++ simulation tools, which shall be updated very soon
TO DO:
- implement a function computing kth-order approximation of simulated moments of y

— allow the user to override compilation flags for MATLAB MEX (it was already
  working for the preprocessor, the MEX for Octave and Dynare++)
— increase the symmetry of MEX build infrastructure between MATLAB and Octave
— when linking MEX for Octave, do not add the output of “mkoctfile -p FLIBS”.
  It is unneeded, and it can create a conflict between the system compiler and
  a user-supplied compiler

By the way:
— restore optimization on macOS for C/C++ MEX (it had been removed in
  5df2392a)
— remove -fno-omit-frame-pointer on MATLAB/Linux, since it would be cancelled
  by subsequent -O2 and should not be needed anyways
— remove FFLAGS under Octave, unused

The current Octave support is utterly broken (both in stable and unstable), it
crashes Octave. It relies on an unofficial Octave app for
macOS (https://octave-app.org), which is infrequently updated.

This commits drop support for Octave in the macOS package. We will now tell our
macOS+octave users to use the Homebrew Dynare package (which is maintained by
the Homebrew team, and is in reasonably good shape).

– remove clang support
– we no longer support 32-bit
– simplify Darwin specific hacks for Octave MEX files

(cherry picked from commit bd97fdc1)

Closes: #1750

— FCFLAGS needs to contain the compilation flags used when compiling Octave,
  otherwise it fails at configure stage when looking for gfortran
— Explicitly add -L$OCTLIBDIR, because with Octave 5 it is no longer there, and
  on Fedora the Octave libraries are no in the default linker search path

This is a Fortran 2008 rewrite of disclyap_fast.m.

Closes: #1737

If something was already in LDFLAGS, it was concatenated without a whitespace,
leading to failure.

By the same token, remove unused support for Fortran 77.

macOS packaging: compile against the version of Octave distributed via https://octave-app.org (currently 4.4.1)

This MEX solves nonlinear systems of equations using a trust region algorithm.
The problem is subdivided in smaller problems by doing a block
triangularisation of the Jacobian at the guess value, using the
Dulmage-Mendelsohn algorithm.

The interface of the MEX is simply:

  [x, info] = block_trust_region(f, guess_value);

Where f is either a function handle or a string designating a function.
f must take one argument (the evaluation point), and return either one or two
arguments (the residuals and, optionally, the Jacobian).

On success, info=0; on failure, info=1.

By the same token, remove unused support for Fortran 77.

The changes in 8065e9d4 were not working as
intended, because AC_CHECK_PROG expect values and not actions. Hence
AC_MSG_ERROR was not properly executed.

By the way, simplify some test conditions using && instead of if/then/fi
blocks.

In particular, if either MATLAB or Octave is missing, one needs to pass either
--disable-matlab or --disable-octave.

Moreover, several new configure flags have been introduced for disabling some
components:
--disable-doc
--disable-dynare++
--disable-mex-dynare++
--disable-mex-ms-sbvar
--disable-mex-kalman-steady-state

This means we now require GCC 8.

Boost filesystem library is no longer needed.

Ref preprocessor#16

As a consequence, remove the --enable-openmp option of the configure script.

It constructs the stacked residuals and jacobian of the perfect foresight
problem.

It is an almost perfect replacement for the perfect_foresight_problem.m
routine, while being much more efficient.

Note however that the DLL never return complex numbers (it instead puts NaNs at
the place where there would have been complex). This may create problems for
some MOD files; the algorithms will need to be adapted to use a more
line-search method.

Adapt the code accordingly.

On Windows, this means that a POSIX threads implementation is no longer needed,
since C++11 threads are implemented using native Windows threads.

On GNU/Linux and macOS, POSIX threads are still used under the hood.

A new m4 macro (AX_CXX11_THREAD) is used to add the proper compilation
flags (instead of AX_PTHREAD).

Needed because bytecode DLL uses some preprocessor code which is C++14.

It's actually not needed. The QZ complex decomposition can be obtained in
Octave on real matrices by casting them to the complex type.

These functions are all included in Octave 4.2.