Building simulator on W10 about epsilon

[MarcNumworks]

Thank you for your issue, amidarius, we have made some modifications similar to what you did in the version 22 which was just released.
The Windows build and CI should be working now.
Building a custom userland for the device (with arm-none-eabi-gcc installed via msys) should work out of the box too.

[ljoets]

Hi,

Thanks @amidarius for your notes !

Just one thing : I needed to type export SETUPTOOLS_USE_DISTUTILS=stdlib (without the set) before pip3 install lz4 pypng stringcase. @MarcNumworks : Is it possible to add this line on the https://www.numworks.com/engineering/software/build/ page ?

So I can now build epsilon.exe, but I noticed several warnings during windows compilation :

CXX     apps/regression/model/trigonometric_model.o
apps/regression/model/trigonometric_model.cpp: In member function 'virtual void Regression::TrigonometricModel::specializedInitCoefficientsForFit(double*, double, Regression::Store*, int) const':
apps/regression/model/trigonometric_model.cpp:194:34: warning: 'xMin' may be used uninitialized [-Wmaybe-uninitialized]
  194 |   double period = 2.0 * std::fabs(xMax - xMin);
      |                         ~~~~~~~~~^~~~~~~~~~~~~
apps/regression/model/trigonometric_model.cpp:187:10: note: 'xMin' was declared here
  187 |   double xMin, xMax, yMin, yMax;
      |          ^~~~
apps/regression/model/trigonometric_model.cpp:194:34: warning: 'xMax' may be used uninitialized [-Wmaybe-uninitialized]
  194 |   double period = 2.0 * std::fabs(xMax - xMin);
      |                         ~~~~~~~~~^~~~~~~~~~~~~
apps/regression/model/trigonometric_model.cpp:187:16: note: 'xMax' was declared here
  187 |   double xMin, xMax, yMin, yMax;
      |                ^~~~
apps/regression/model/trigonometric_model.cpp:190:32: warning: 'yMin' may be used uninitialized [-Wmaybe-uninitialized]
  190 |   modelCoefficients[0] = (yMax - yMin) / 2.0;
      |                          ~~~~~~^~~~~~~
apps/regression/model/trigonometric_model.cpp:187:22: note: 'yMin' was declared here
  187 |   double xMin, xMax, yMin, yMax;
      |                      ^~~~
apps/regression/model/trigonometric_model.cpp:190:32: warning: 'yMax' may be used uninitialized [-Wmaybe-uninitialized]
  190 |   modelCoefficients[0] = (yMax - yMin) / 2.0;
      |                          ~~~~~~^~~~~~~
apps/regression/model/trigonometric_model.cpp:187:28: note: 'yMax' was declared here
  187 |   double xMin, xMax, yMin, yMax;
      |                            ^~~~

CXX     eadk/src/simulator.o
eadk/src/simulator.cpp: In function 'KDPoint p(eadk_point_t)':
eadk/src/simulator.cpp:50:11: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
   50 |   return *reinterpret_cast<KDPoint *>(&point);
      |           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
eadk/src/simulator.cpp:50:11: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
eadk/src/simulator.cpp: In function 'KDRect r(eadk_rect_t)':
eadk/src/simulator.cpp:55:11: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
   55 |   return *reinterpret_cast<KDRect *>(&rect);
      |           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
eadk/src/simulator.cpp:55:11: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
eadk/src/simulator.cpp: At global scope:
eadk/src/simulator.cpp:95:15: warning: 'eadk_external_data_size' initialized and declared 'extern'
   95 | extern size_t eadk_external_data_size = 0;
      |               ^~~~~~~~~~~~~~~~~~~~~~~

CXX     poincare/src/list.o
poincare/src/list.cpp: In member function 'Poincare::Expression Poincare::List::approximateAndRemoveUndefAndSort(const Poincare::ApproximationContext&) const [with T = float]':
poincare/src/list.cpp:233:42: warning: array subscript -4 is outside array bounds of 'Poincare::Evaluation<float> [1]' [-Warray-bounds=]
  233 |   static_cast<ListComplex<T>&>(eval).sort();
      |   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~
poincare/src/list.cpp:228:17: note: at offset -8 into object 'eval' of size 2
  228 |   Evaluation<T> eval =
      |                 ^~~~
poincare/src/list.cpp: In member function 'Poincare::Expression Poincare::List::approximateAndRemoveUndefAndSort(const Poincare::ApproximationContext&) const [with T = double]':
poincare/src/list.cpp:233:42: warning: array subscript -4 is outside array bounds of 'Poincare::Evaluation<double> [1]' [-Warray-bounds=]
  233 |   static_cast<ListComplex<T>&>(eval).sort();
      |   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~
poincare/src/list.cpp:228:17: note: at offset -8 into object 'eval' of size 2
  228 |   Evaluation<T> eval =
      |                 ^~~~

CXX     poincare/src/list_sort.o
poincare/src/list_sort.cpp: In member function 'Poincare::Expression Poincare::ListSort::shallowReduce(Poincare::ReductionContext)':
poincare/src/list_sort.cpp:73:27: warning: array subscript -4 is outside array bounds of 'Poincare::Evaluation<float> [1]' [-Warray-bounds=]
   73 |       listOfDefinedScalars};
      |                           ^
poincare/src/list_sort.cpp:63:21: note: at offset -8 into object 'approximatedList' of size 2
   63 |   Evaluation<float> approximatedList =
      |                     ^~~~~~~~~~~~~~~~

CC      python/port/stackctrl.o
python/port/stackctrl.c: In function 'mp_stack_ctrl_init':
python/port/stackctrl.c:36:32: warning: storing the address of local variable 'stack_dummy' in 'mp_state_ctx.thread.stack_top' [-Wdangling-pointer=]
   36 |     MP_STATE_THREAD(stack_top) = (char*)&stack_dummy;
python/port/stackctrl.c:35:18: note: 'stack_dummy' declared here
   35 |     volatile int stack_dummy;
      |                  ^~~~~~~~~~~
In file included from python/port/stackctrl.c:27:
python/src/py/mpstate.h:287:23: note: 'mp_state_ctx' declared here
  287 | extern mp_state_ctx_t mp_state_ctx;
      |                       ^~~~~~~~~~~~

[MarcNumworks]

Hi @ljoets,

Thanks for reporting these warnings, they don’t appear with clang or arm gcc but indeed they also appear with gcc on linux, we will fix them.

I have added your question in an internal issue where we want to improve this tutorial, I hope we will have time for this soon.

Have a nice day !

[ljoets]

Here the warnings when actually building on Debian 12 :

XX     poincare/src/list.o
poincare/src/list.cpp: In member function ‘Poincare::Expression Poincare::List::approximateAndRemoveUndefAndSort(const Poincare::ApproximationContext&) const [with T = float]’:
poincare/src/list.cpp:233:42: warning: array subscript -4 is outside array bounds of ‘Poincare::Evaluation<float> [1]’ [-Warray-bounds]
  233 |   static_cast<ListComplex<T>&>(eval).sort();
      |   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~
poincare/src/list.cpp:228:17: note: at offset -8 into object ‘eval’ of size 2
  228 |   Evaluation<T> eval =
      |                 ^~~~
poincare/src/list.cpp: In member function ‘Poincare::Expression Poincare::List::approximateAndRemoveUndefAndSort(const Poincare::ApproximationContext&) const [with T = double]’:
poincare/src/list.cpp:233:42: warning: array subscript -4 is outside array bounds of ‘Poincare::Evaluation<double> [1]’ [-Warray-bounds]
  233 |   static_cast<ListComplex<T>&>(eval).sort();
      |   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~
poincare/src/list.cpp:228:17: note: at offset -8 into object ‘eval’ of size 2
  228 |   Evaluation<T> eval =
      |                 ^~~~

CXX     poincare/src/list_sort.o
poincare/src/list_sort.cpp: In member function ‘Poincare::Expression Poincare::ListSort::shallowReduce(Poincare::ReductionContext)’:
poincare/src/list_sort.cpp:73:27: warning: array subscript -4 is outside array bounds of ‘Poincare::Evaluation<float> [1]’ [-Warray-bounds]
   73 |       listOfDefinedScalars};
      |                           ^
poincare/src/list_sort.cpp:63:21: note: at offset -8 into object ‘approximatedList’ of size 2
   63 |   Evaluation<float> approximatedList =
      |                     ^~~~~~~~~~~~~~~~

LD      epsilon.bin
/usr/bin/ld: attention: output/release/simulator/linux/ion/src/simulator/shared/collect_registers_x86_64.o: l'absence de la section .note.GNU-stack implique une pile exécutable
/usr/bin/ld: NOTE: Ce comportement est déprécié et sera supprimé dans une version ultérieure de l'éditeur de liens

The build bin file works fine.