[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [PATCH 1/5] Use a function to add a device to the partition gui.



On Wed, 2009-12-09 at 16:58 +0100, Hans de Goede wrote:

> >>
> >> This is wrong, now handleUdevDiskLabelFormat() will get called on them
> >> which in case of a multipath or a RAID mirror will succeed (and in
> >> case of a RAID stripe might succeed too), resulting in the raw disk
> >> becoming part of storage.partitioned. Maybe partionable should
> >> become a parameter of the DiskDevice constructor and get set to
> >> False here ?
> >
> > Instantiating DiskDevice for a disk is most certainly not wrong.
> 
> Sorry I did not mean to imply the principle was wrong, merely that as
> implemented it will result in a wrong outcome in some cases.

I'm a little cranky in the morning. I'd have been able to tell if I'd
had some coffee first. No worries.

> >> Even with this fixed, the raw disks now will still become part of
> >> storage.disks, and I'm not sure we even want that.
> >
> > I'm not sure why we wouldn't now that the (disk ==>  disklabel) rule is
> > gone. However, we could certainly use those formats' new "hidden"
> > attribute for this if we do want to filter them out any place they are
> > found to be problematic.
> >
> 
> Ah, I forgot about the hidden attribute, and I agree that in general
> making these disks DiskDevices is the right thing to do. All I was
> trying to say is that I'm worried this might cause issues.

I ended up using the "hidden" attribute as an indicator not to do
disklabel handling for these disks. New patches coming soon...

> 
> (Likewise for the new status behavior of DMRaidArrayDevice / MultiPath...)

This really should work fine, and it's absolutely the more correct
approach, so if there are problems I think they'll be problems that need
addressing anyhow.

Dave


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]