Piotr Bartman-Szwarc

Interesting indeed. I don't see anything against proposed change, so we can do it

I can change the red badge to green, no problem. I also fixed the issue with the extra steps. However, changing the message content isn't entirely a good idea. The...

Ok, so I need a little more clarification what should be done. My point is: 1. We should never assure a user that something is up-to-date if there is still...

Yes, definitely. The current implementation of `shuffle_local` seems to be buggy and we don't have a uniform distribution there.

@k4z4n0v4 thanks for contribution! For me PR looks nice. @marmarek I think it should be merged after #141

Ok, I'm writing the docs > * you use vendor/product as strings from hwdata - are they stored anywhere (or otherwise expected to not change)? in rare cases they may...

> > > * why only PCI devices can be "required"? > > > > I think we've decided that there is no benefit in making USB/block `required`. > >...

> Would this be a good opportunity to use diskseq to identify block devices? diskseq + partition number is guaranteed to be unique until the backend is rebooted, whereas other...

> few comments on the recent push; I'm waiting with the full review for the documentation. Most parts of docstring in `qubes/devices.py` are the same since the main concept remains...

> Since the API now considers also `self_identity` when automatically attaching devices (which is good!), I think there need to be support for multiple assignments of the same port but...