1 of 1 people found this helpful
You can get the cross compile toolchain from the repository, (here Edison SW Downloads),
or you can generate it from the source archive with the bitbake option -c populatesdk.
Then, to install it, run the resulting bash script.
Finally, to use the tool chain, you have many possibilities:
#1) You use it "as is", sourcing the environment and running the appropriate gcc/g++ command.
From slide #17 (NB: this was for Galileo, but it's very similar): IntelAcademic_IoT_07
#2) You can chroot the root folder (the one in /opt/edison/...) and compile your program just like if you were on Edison (but with your computer's resources).
#3) For developing integrated project, you should use an IDE like you suggested (Eclipse, Codeblock...) . You've to set it up to use the appropriate compiler.
It's also possible to integrate a tool in the IDE to send your binaries to Edison via SSH.
Take a look on page 18 for some clues: Intel-DeveKit-Tutorial-Guide.pdf
There should be no Edison specific settings as the entire tool chain has been specifically built for Edison.
Hope this helps.
I'd like to expand on this question a little.
I'm attempting to build a kernel module from within the chroot environment that you mentioned in your answer #2. However, the build fails with the make complaining that there are missing kernel paths because it is using uname to locate build headers. I realize this can be workaround by providing the path in place of uname within the makefile, but then the the header files are still missing. So how to generate the header files?