Can developers for Android use 1.4 in a serious manner, or is it better for them to work on tip? Or should we just supply a set of patches, or a branch, on top of 1.4? My understanding was that the Android 1.4 support was quite experimental. It doesn't seem like a good idea if we're going to have a set of Android patches for 1.4.2 also.

I suspect the compiler transition will make it hard to use tip for quite a while.

I would be happy with an android-1.4 branch. As is, 1.4 is reasonably stable for GOOS=android, it is the mobile repository (and build system) that is in flux.

I'll put this in 1.4.1, because it seems less terrible than having a separate android-1.4 branch. I hope this will be the end though. (And I'll get the Copyright fix.)

The code review for this cherry-pick suggested that /dev/log/main doesn't exist on Android L devices. https://go-review.googlesource.com/2803

Based on that and the fact that the malloc change doesn't apply cleanly at all (the change is in a Go file that is still written in C in the release branch), I suggest that you either have people live with the 1.4 limitations or publish a github.com/crawshaw/go-android for people to use. It's too much work (and too uncertain to be useful) to get these into the release branch, and I don't want to create yet another official branch in the main repo that will be maintained.

I don't consider the Android L logging problem to be important. L devices remain extremely rare and so app writers are not writing apps against it. But I will go with your suggestion of creating a copy elsewhere.