What are best practices using the MATLAB Embedded Coder with Version Control (git)?

I'm trying to setup a project to generate C code from a Simulink model. The generated code will be used by external existing code. Git is used for version control.

The Embedded Coder generates a lot of files additional to the "relevant" .c and .h files like:

• buildInfo.mat
• codedescriptor.dmr
• codeInfo.mat
• compileInfo.mat
• defines.txt
• modelsources.txt
• rtw_proj.tmw
• trwtypeschksum.mat
• ...

Most of these files are binary and there seems to be no benefit in tracking them with git.

I already figured out that "View Code" in Simulink is not possible when cloning a repository without the buildInfo.mat. So this seems to be a file that should be tracked by git.

What's the benefit in tracking the other files?

Is there a "best practice" on which files should be included in version control?

Which files are necessary that someone else can continue working with the embedded coder after cloning my remote repository?

Thanks for any advice :)

[Dennis Hirth]

I would suggest to only version the .c and .h files in git. All other files can be reproduced.