OK, considering only the technical aspects of this bug (not that the other aspects are not important but I don't want to take position on these), I'm ready to prepare a NMU. Before doing so, I'd like everyone's point of view on the solution I proposed as it introduces a regression for slow to appear devices. Given the fact that this bug defeats the purpose of raid 1, I think the regression is worth it in the first time (for sure, we need a better solution in the long term). What do you think about it?
Cheers Yann