To make integration of new upstream versions as easy as possible, I like to isolate my own work from the Intel releases as much as possible.
To that end, I work from a "local" Yocto layer, and I have defined my own image, based on the default edison-image.
Unfortunately, the postBuild and flashall scripts assume that the image produced by bitbake will be named "edison-image", and postBuild is always pointed at the same directory (.../edison_src/build/toFlash).
It seems useful (and straightforward) to be able to manage a set of images by adding a directory layer below toFlash/. Also, in the interest of keeping images around, even if one wishes to torch the whole build directory, moving the toFlash directory down a level seems like a good idea to me. This would also enable releases come with a pre-built default image (named, e.g., edison-default-image) that wouldn't have to be built locally, and wouldn't normally get overwritten.
So I am attaching the scripts with the modifications that I have made in hopes that some or all of it might get pulled upstream.
The postBuild.sh script now uses argument 1 to identify an image name other than "edison-image" (previously, this could override the build directory?). I think that a better solution would use some output from bitbake to identify the most recently built image, but I haven't looked into that yet.
The flashall.sh script now uses the name of the directory it is in to identify the .hddimage and .ext4 files, since the build process names them that way.
I have also included @David_J_Hunt sudo check at the beginning of flashall.sh from this discussion: flashall.sh changes to allow re-partitioning without xFSTK. (This still burns me as often as not!)