Question about generating an executable with multiple source files

[jinz2014]

Suppose an executable depends on multiple source files, and one of the source files contains CUDA APIs. Can you please show the command to generate an executable using cocl_py ? People may prefer not to include source files using "#include ".

Thanks

[hughperkins]

Yeah, I mean, you just use a standard linker, ie g++. The test/endtoend/multi is an example, but lacks a build script for some reason. A example of linker commands though is at https://github.com/hughperkins/coriander/blob/master/test/endtoend/sharedlibrary/build.sh, though this example is a bit of a poor role model, since it's actually creating a shared library from one file, and then linking the second file with this shared library. In fact, to link multiple files, simply pass all the .o files into a single g++ command.

…

On Sun, Jul 26, 2020, 19:57 jinz2014 ***@***.***> wrote: Suppose an executable depends on multiple source files, and one of the source files contains CUDA APIs. Can you please show the command to generate an executable using cocl_py ? People may prefer not to include source files using "#include ". Thanks — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#94>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAA6FKHLOYSPDU7WXCRUEETR5S7EZANCNFSM4PIHCKKQ> .