It occurs to me that it might be most expeditious in the first release to target only what I call "posix" - i.e., linux + BSD & similar - this also gives you OSX-command-line and maybe Windows-plus-Cygwin-command-line. Maybe also limit it to file-container-only including hidden volume within that container.
My thinking is that pinning down the toolchain(s) for the initial targets might be more straightforward this way. By using autotools + maybe gcc (or clang), it seems to be that you could rapidly come to a non controversial solution.
It's completely beside the point that I personally couldn't give a fork about Windows ever. I swear it is
It could easily be that the prospect is just too disgusting to too many people, and that it is felt it would give a bad impression of the project from the beginning, or that it really wouldn't save enough time-to-initial-release to be significant in the grand scheme of things.