Hi everyone, I have a python code that internally uses *numpy *(plus other dependencies) to perform some basic numerical operations.
I installed *numpy* as "*conda install numpy*" (Windows 10) meaning that it includes and uses the *mkl libraries*. After running pyinstaller on this code, I have an executable that contains the *mkl libraries. *I can see those dlls when I run the executable and check the folder created in Temp with all the dependencies uncompressed. Nevertheless, when I run the executable, t*he program crashes* when it tries to use numpy with the optimized mkl libraries (a *numpy.dot*(..,..) operation). I copied the executable to the folder where I have the dlls of the mkl libraries in my system and the program runs properly. *Question*: why the mkl libraries are not properly detected/used during runtime even if they are included in the executable? Is there a way to set the .specs to link properly to those dlls instead of copying always the dlls next to the program or adding to the path? If not, is there a way to properly exclude the mkl libs from the executable if in any case I need to use them from a different folder ? Thanks, Ricardo RaC -- You received this message because you are subscribed to the Google Groups "PyInstaller" group. To unsubscribe from this group and stop receiving emails from it, send an email to pyinstaller+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/pyinstaller/20a0f688-4032-409c-b48f-3e777a32cb57o%40googlegroups.com.
