Flint installation error: make: *** [build/generic_files/memory_manager.lo] Error 1

Question

I'm trying to install v3.1.2 of flint into my account on the MPCDF HPC cluster. I managed to install GMP 6.3.0, MPFR 4.2.1 in my home directory with ./configure --prefix=$HOME/software/<packagename> for the two packages.

From the flint directory,I followed the online docs and manage to get till the ./configure step using:

>>> cd flint/
>>> ./bootstrap.sh
>>> ./configure --prefix=$HOME/software/flint/ --with-gmp-include=$HOME/software/gmp-6.3.0/include/ --with-gmp-lib=$HOME/software/gmp-6.3.0/lib/ --with-mpfr-include=$HOME/software/mpfr-4.2.1/include/ --with-mpfr-lib=$HOME/software/mpfr-4.2.1/lib/

However, when I move to the make step, I get this output:

/usr/bin/mkdir -p build/generic_files
  CC  generic_files/inlines.c
  CC  generic_files/profiler.c
  CC  generic_files/sprintf.c
  CC  generic_files/fscanf.c
  CC  generic_files/memory_manager.c
In file included from ./src/thread_pool.h:15:0,
                 from src/generic_files/memory_manager.c:18:
./src/flint.h:154:28: error: unknown type name ‘_Thread_local’
 #  define FLINT_TLS_PREFIX _Thread_local
                            ^
src/generic_files/memory_manager.c:340:1: note: in expansion of macro ‘FLINT_TLS_PREFIX’
 FLINT_TLS_PREFIX size_t flint_num_cleanup_functions = 0;
 ^
src/generic_files/memory_manager.c:340:25: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘flint_num_cleanup_functions’
 FLINT_TLS_PREFIX size_t flint_num_cleanup_functions = 0;
                         ^
In file included from ./src/thread_pool.h:15:0,
                 from src/generic_files/memory_manager.c:18:
./src/flint.h:154:28: error: unknown type name ‘_Thread_local’
 #  define FLINT_TLS_PREFIX _Thread_local
                            ^
src/generic_files/memory_manager.c:342:1: note: in expansion of macro ‘FLINT_TLS_PREFIX’
 FLINT_TLS_PREFIX flint_cleanup_function_t * flint_cleanup_functions = NULL;
 ^
src/generic_files/memory_manager.c:342:43: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘*’ token
 FLINT_TLS_PREFIX flint_cleanup_function_t * flint_cleanup_functions = NULL;
                                           ^
src/generic_files/memory_manager.c: In function ‘flint_register_cleanup_function’:
src/generic_files/memory_manager.c:358:5: error: ‘flint_cleanup_functions’ undeclared (first use in this function)
     flint_cleanup_functions = flint_realloc(flint_cleanup_functions,
     ^
src/generic_files/memory_manager.c:358:5: note: each undeclared identifier is reported only once for each function it appears in
src/generic_files/memory_manager.c:359:10: error: ‘flint_num_cleanup_functions’ undeclared (first use in this function)
         (flint_num_cleanup_functions + 1) * sizeof(flint_cleanup_function_t));
          ^
src/generic_files/memory_manager.c: In function ‘_flint_cleanup’:
src/generic_files/memory_manager.c:380:21: error: ‘flint_num_cleanup_functions’ undeclared (first use in this function)
     for (i = 0; i < flint_num_cleanup_functions; i++)
                     ^
src/generic_files/memory_manager.c:381:9: error: ‘flint_cleanup_functions’ undeclared (first use in this function)
         flint_cleanup_functions[i]();
         ^
src/generic_files/memory_manager.c: At top level:
cc1: warning: unrecognized command line option "-Wno-stringop-overread" [enabled by default]
cc1: warning: unrecognized command line option "-Wno-stringop-overflow" [enabled by default]
Makefile:615: recipe for target 'build/generic_files/memory_manager.lo' failed
make: *** [build/generic_files/memory_manager.lo] Error 1

Can someone help me figure out where the problem is?

Answer 1

For now the solution seemed to be to use an older release v3.0.1. Running through the same steps stated above lead to a succesful installation. I did do one thing slightly different - which is to load the OpenMP module, but I'm not sure if this makes any difference. For now I'm happily able to use python-flint after installing v3.0.1

Details - following accepted answer above

---

The key factor was indeed the gcc version in play as suggested by John Bollinger. The default gcc is version 4 point something, which is already very old (the last v4 release, 4.8.5 was back in 2015).

Once I used gcc-13 with module load gcc/13 and ran make, the process continued as expected (without the need to alter the configure step.