Bazel build support for Windows [Probably blocked: #3024]

[rt4914]

Currently the project does not work with bazel-build on Windows operating system. This needs to be fixed considering the contributors are generally students with Windows operating system and they are willing to contribute but this restricts there contribution to minor code changes.

Also this is really important from next year's GSoC point of view too before which we see a good number of contributors.

[rt4914]

cc: @fsharpasharp

[BenHenning]

@rt4914 do you have context on what's going on here? We know that the Android Studio plugin may not work in Windows but the build itself should work in theory. What problems have contributors reported having?

[BenHenning]

Hmm. Trying to build locally and I ran into bazel-contrib/rules_jvm_external#288. Unfortunately, that issue was closed without a clear resolution, so it may be the case that the build won't work locally with native windows. I'm going to keep poking around, but the alternative might be to use a Unix subsystem on Windows or a VM.

[BenHenning]

So I've finally gotten a Windows build working. Double checking that Robolectric & JUnit tests pass. Unfortunately, it requires using a Unix subsystem instead of native Windows--I suspect we'll need to work with the Bazel team to try and get the Windows build working. This should be a reasonable stop-gap, though having Windows Android Studio installed with Ubuntu-based Bazel for building may be confusing. This is probably best as a stop-gap until we can get fixes in the build pipeline for Windows.

[BenHenning]

https://github.com/oppia/oppia-android/wiki/Bazel-Setup-Instructions-for-Windows is now live. @rt4914 can you verify with the contributors whether these instructions help unblock them?

[BenHenning]

Took a while, but confirming that all command line-compatible tests seem to pass on the subsystem.

[rt4914]

@BenHenning I will inform about this to the contributors.

[BenHenning]

I have a slight suspicion that we need to resolve #3024 before this since the above issue & PR linked from that issue implied that there may be an aapt2 incompatibility that gets resolved in 30.+.