Why wouldn't getting as close to 100% compatibility as possible be the aim? WSL is currently beta. The installation documentation has a big warning at the top that functionality is currently incomplete.
All I can say is that the first two things I tried on the Windows Linux Subsystem didn't work, mainly running Elixir and Debootstrap. It's okay that they aren't (yet?) supported, but something that looks & smells like Ubuntu, yet isn't, doesn't inspire confidence that it can be trusted to act like Ubuntu.
An alternative approach would be to call it Winbuntu, Winux, or simply Ubuntu for <windows subsystem architecture>. Then aggressively maintain a list of known compatibility issues, and stay on top of fixing them. Maybe even open source it?! A close sourced, not-quite-what-I-wanted solution is simply less interesting to me.
I'm not sure a public-facing curated list of specific compatibility issues would be practical at this stage of the project's life; it's changing fairly rapidly. FWIW, the bug tracker is public, and the team seems to be fairly responsive; a UPX Compression bug I ran into was tagged fixinbound within a couple weeks. https://github.com/Microsoft/BashOnWindows
The speed at which they can push new builds to people who aren't on Windows Insider is a bit unfortunate, though.
> All I can say is that the first two things I tried on the Windows Linux Subsystem didn't work, mainly running Elixir and Debootstrap. It's okay that they aren't (yet?) supported, but something that looks & smells like Ubuntu, yet isn't, doesn't inspire confidence that it can be trusted to act like Ubuntu.
You need to handle this anyways, as it's unlikely that your target deploy platform will always be ubuntu. In any case, you can just use docker like most OSX folks do now.
I'm not familiar with Elixir or Debootstrap (I'm guessing web-oriented)? But, I've had great success with using WSL for C++ oriented projects, building with both g++ & clang. Only issue in that regard I've had is that the emulated syscalls cannot currently enumerate ethernet devices. The calls are implemented, but fail at runtime currently. I've not yet filed a bug report, because I don't know how it should perform.
I don't think they can open source WSL in any useful capacity because it's signed as part of the kernel itself - afaict this is also why an update to WSL is an update to the Windows build itself.
