what lesson do you believe they learnt, and why do you think it was the wrong one? they clearly noted that (a) the gecko component was hard and fiddly to embed and (b) fixing it would draw away scarce resources from more critical areas, so they decided to not waste both their and their users' time, and remove embedding support. i think it was the right way to go; as kibwen noted, servo can be designed around embedding support right from the beginning.