Compiling Dynare error: unknown type name ‘char16_t’

[lombard]

Hi,
I've tried to install Dynare from repository in Ubuntu 16.04 LTS, as well as compiling it, but I always get this error message

error: unknown type name ‘char16_t’
typedef char16_t mxChar;

I'm using Matlab 2010a. In the past I've never faced this problem (I think).

I've tried also with Fedora 25, with similar results.

Does anybody know a solution?

Thanks

[jpfeifer]

Hi, where exactly does this error appear? To check the status of the problem, see https://github.com/DynareTeam/dynare/issues/1344

[MichelJuillard]

Hi Gianni,

can you tell us what is the version of your compiler:

gcc --version

We don't have access to Ubuntu machines and can't test it directly.

Best,

Michel

[lombard]

Hi Michel,
Good to hear from you, and thanks for picking up this help message.
I encountered the same problem in Fedora.
... meanwhile I managed to fix it at least under Ubuntu. The fix consists of replacing the Matlab headers (in external/include) of Matlab 2010, with newer versions.
Then the Ubuntu binary (.deb) worked out of the box.

I'm now giving a try with Fedora. You will hear from me if it doesn't work
Best,
Gianni

[lombard]

Hi, and here I'm again.
Compiling from source is not as easy (as it used to be)
Here is what I get when I run dynare after compilation

[mex] Generalized QZ.
[m] Sylvester equation solution.
[mex] Kronecker products.
[mex] Sparse kronecker products.
[m] Local state space iteration (second order).
[mex] Bytecode evaluation.
[no] k-order perturbation solver.
[no] k-order solution simulation.
[no] Quasi Monte-Carlo sequence (Sobol).
[no] Markov Switching SBVAR.

Using 64-bit preprocessor
"/home/gianni/gitdynare/dynare/matlab/preprocessor64/dynare_m" bkk.mod: Aborted
Starting Dynare (version 4.5-unstable).
Starting preprocessing of the model file ...
Substitution of endo leads >= 2: added 9 auxiliary variables and equations.
Substitution of endo lags >= 2: added 10 auxiliary variables and equations.
Found 42 equation(s).
Evaluating expressions...done
terminate called after throwing an instance of 'SymbolTable::SearchFailedException'

Note that a number of mex were not compiled and it failed to compile dynare++ as well

Configuration went through ok. The only missing package, I guess, is slicot, which I couldn't find anywhere for fedora (.rpm)

Best

Gianni

[lombard]

And I'm using

g++ --version
g++ (GCC) 6.2.1 20160916 (Red Hat 6.2.1-2)

[jpfeifer]

Hi,
1. Are all the required packages mentioned at https://github.com/DynareTeam/dynare installed?
2. What exact command did you use for

Code: Select all

./configure

, i.e. which options did you invoke?
3. Could you please provide the output from the configure and make commands.

[lombard]

Hi,
Thanks!
I'm calling with

./configure --with-matio=/usr/lib/ --with-matlab=/opt/matlab10/ MATLAB_VERSION=10.1

I have all the packages up to slicot (I've just got the scripts from SLICOT but not compiled yet)

the output of config attached in the text file

Gianni

[jpfeifer]

The config-file looks good. You don't necessarily need SLICOT, only for the steady state Kalman filter. What is puzzling is that many mex-files that should be compiled were not created, like e.g.

Code: Select all

k-order perturbation solver

When you run

Code: Select all

make all

, are there any error messages related to these files?

P.S: I am not sure it makes a difference, but your Matlab version cannot be 10.1. Releaste 2016b has version number 9.1.

[lombard]

Hi,
Thanks. The version was a typo (10.1 for 2010a).
I've now upgraded to 2015a. Things work up to a point: same missing mex functions

[mex] Generalized QZ.
[m] Sylvester equation solution.
[mex] Kronecker products.
[mex] Sparse kronecker products.
[m] Local state space iteration (second order).
[mex] Bytecode evaluation.
[no] k-order perturbation solver.
[no] k-order solution simulation.
[no] Quasi Monte-Carlo sequence (Sobol).
[no] Markov Switching SBVAR.

The error message is indeed related to dynare++, I think

make[2]: Entering directory '/home/gianni/gitdynare/dynare/mex/build/matlab/libdynare++'
make[2]: *** No rule to make target '../../../../dynare++/kord/faa_di_bruno.cpp', needed by 'libdynare___a-faa_di_bruno.o'. Stop.
make[2]: Leaving directory '/home/gianni/gitdynare/dynare/mex/build/matlab/libdynare++'
Makefile:408: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/home/gianni/gitdynare/dynare/mex/build/matlab'
Makefile:425: recipe for target 'all-recursive' failed
make: *** [all-recursive] Error 1

[lombard]

Hi,
The problem was with Dynare from git. I've downloaded the zip file with the scripts directly (not through git as described in the instructions) and everything works fine.
Thanks
Gianni

[HoutanBastani]

Hi,

In the root Dynare directory, please type:

Code: Select all

$ make distclean
$ autoreconf -si
$ ./configure <<with your options>>
$ make


Please post all of the output of each of these commands.

[lombard]

Hi,
I've recently upgraded to master-2017-03-16-1022e67 (Matlab 2016b on Fedora 24 )
and got this error

Function 'subsindex' is not defined for values of class 'struct'.
Error in GB_subsidy (line 1102)
oo_.dr.eigval = check(M_,options_,oo_);
Error in dynare (line 223)

Dynare works fine if I comment out "check;" in the mod file.

Best,
Gianni

[jpfeifer]

Hi Gianni,
what is line 1102 of GB_subsidy.m where the error appears?

[lombard]

Hi Johannes,
it's

oo_.dr.eigval = check(M_,options_,oo_);

Are the inputs in check still expected to be structures in the new unstable Dynare?

Gianni
PS: sorry for the subject of the thread. I've changed it. I don't know why the old one was sticky.