It sounds like you did not take that choice lightly, then, but thoroughly considered it, after all. Yet, I'm surprised that things like power-management had to be rewritten by you guys.
I'm generally trying to resist the temptation to blame hardware designers for everything, but (whatever, here we go: h/w designers are the most natural culprit in any case) it sounds much like the vanilla kernel's power management did not work because the components of the BOOX did not abide to those specifications which are implemented by the kernel.
Why else did the vanilla kernel with all its schedulers, powersave options etc. not suffice? On the most basic level, I hope that the only thing which remained to be done is write a
framebuffer module for the e-ink interface. Without any further modifications, you could then directly use GTK+ and all its applications or just write very light applications with DirectFB. Toolkits like GTK+ would even transparently handle redrawing operations such that only the region which needs update is updated (for everything else, an according optimization of the FB should be trivial).
After all, the only reason for which you'd have had such enormous problems I can think of is that most of the h/w components were not supported by the linux kernel but only in the android tree. I certainly understand, though, that ONYX can't but choose from a limited range of components for the ereader market. Shame on those hardware designers
