If the upstream response is “we don’t want that, throw it out and use this other thing” (which it is in some cases), we might not just jump at doing that, especially if it requires a huge effort across the entire Android stack to replace something that already works great for us with something somebody else would prefer we use instead.

We’ve got a lot of work to do and devices to ship and a very small team to do this all with. If mainline says “your code sucks, we don’t want it”, that’s certainly their right, but I then am puzzled about how we become the bad guys here. We’re not refusing to give anybody the code. It’s freely available under GPLv2.

We have no illusion that things can get upstream without some change and some compromise, but from my perspective this has to be a two way street. If nobody wants to even attempt to meet us halfway, we’re looking at far far more work to rewrite code that mainline doesn’t like than to just keep rebasing our patchsets every other kernel version or so (as we have been since 2.6.14).

This post was mentioned on Twitter by jwildeboer: New blog post: WRT FOSDEM Beer Event http://jan.wildeboer.net/2010/02/wrt-fosdem-beer-event/...