Can't compile project in Eclipse due to logic in header files

[kigster]

I am restarting this topic, because I really like this project and would love for it to thrive.

I did some research, and here what I gathered so far:

• Arduino 1.0.6 does some magic with the files in the project directory: I think it concatenates them all into a single .cpp file and then proceeds to compile/link it. This method works for this project.
• Arduino 1.5.+ changed how they build that single .cpp file, it is now called .ino.cpp, and contains includes for the header files, instead of direct inlining.
• Eclipse I am using, is using 1.5 strategies, and can not compile this project (and neither can Arduino 1.5) due to circular references: because all logic and declarations are in header files, compiler can not resolve external symbols and fails every file.

What was the motivation behind putting all the logic into header files, instead of .cpp as is done traditionally? I am relatively new to C++ (but not C) and I was rather surprised when I saw that. I found a couple of interesting answers on stackexchange:

http://stackoverflow.com/questions/583255/c-code-in-header-files

I believe that to fix my issue each .h file needs to be broken off into .h and .cpp. I am willing to try this effort, but if this does end up working, what are the chances of a pull request?

[jasoncoon]

Absolutely, I welcome the help. I haven't tried Arduino 1.5 since it's still technically "beta", and not supported by Teensyduino yet. As soon as it's supported, I'll update. I haven't been able to get Eclipse working with the Teensy yet, but admittedly I haven't spent a lot of time on it.

I've spent most of my professional career using C# and Java. I worked on a few simple Arduino projects before starting Aurora, which is the first C++ project I've created.

Honestly, I didn't bother splitting the code between cpp and header files because I get very little spare time to work on Aurora, and it seemed like a waste of time. I did a bit of reading on it, and the top reasons to split them seemed to mostly apply to creating libraries for use by others, and for faster compilation speed, which doesn't seem to be a concern for a project of this size.

I'm not opposed to splitting the code properly between cpp and header files, especially if it enables you and others to work on it with Eclipse or other tools. It'd be great if you don't mind doing it, and I'll gladly merge your pull request, after reviewing it, of course. :)

Thanks!