I understand that routines for adc, i2c and serial communications etc are individual micro-processors. My work is 100% with Microchip, but I worked few months for Freescale Semiconductors. I also agree that documentation for these routines shouldnít be part of the documentation of the compiler.
I developed an open source project here in Brazil, this project is based on the project Gogo Board. The Gogo Board framework is a collection of open-source hardware platforms mainly aimed for educational projects. Its uses a microchip micro-processor.† I would like very much to use sdcc, but the compiler I currently use has an extensive library. I just canít tell to my boss that we should use sdcc and that all routines must be rewritten from scratch, just to be 100% open source. I would be very happy with this freedom , but I will waste so much time just to get where I'm now. The master Google will help me, but he won't† make my task for me , yet :P .
Why the sdcc donít have a model of implementation for the libraries. For example, i2c library has methods for writing, reading, waiting etc. Thus each micro-processor has its own implementation, and the user can work in a higher level. If I want to that my routines be added in the library must follow the standard model of implementation.
This is what I really want.
The bests regards.