This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

How many Include paths...

In general, how many -I include path specifications should an ARM compile require?

Assume that I want to share a compiler toolchain for Cortex-M, and am programming essentially "bare metal."  (obviously if I had external libraries, they'd each probably require a path.)

I'm used to single-vendor architectures like Atmel AVR, where the device include files are lumped in with the compiler, and you can get by with "#include <avr.io.h>" in your C code with NO include paths in the compile line, because the "avr" directory got merged into the compiler install.   I don't think that scales very well for a large number of separate sub-architectures and vendors (Hmm. maybe it would be ok?)
So I'm thinking

  1. CMSIS Core path.
  2. Vendor path.

(You can think of this as "The RPi Pico structure, where "blink" requires 37 separate include paths on the compiler line, should be reorganized!  But how?")

Individual chip vendors don't seem to be very helpful: "just install OUR toolchain!  A modern development system has plenty of space for multiple copies of the C compiler!  Besides, you don't want to use those other vendors' chips, anyway!!"  Sigh.