[blivet 0/8] Refactor abstract RAID concept out of mdraid.py

Jan Safranek jsafrane at redhat.com
Fri Nov 22 08:17:36 UTC 2013


On 11/21/2013 06:39 PM, David Lehman wrote:
> On Thu, 2013-11-21 at 11:59 -0500, mulhern wrote:
>> These patches support a refactoring of mdraid.py into two files.
>> The new file, raid.py contains a bunch of classes representing RAID
>> levels.
> 
> I've only read over these briefly so far and they look good overall. I
> do have a few initial comments.
> 
>>
>> The first four patches add some unit tests to catch any regressions
>> that the refactoring might introduce. In some cases, some changes
>> were made to the tested code, as seemed appropriate.
>>
>> The next patch actually implements the RAID classes and their tests.
> 
> The only real complaint I have here is the continued use of integers as
> the primary means of identification of RAID levels. I really want part
> of this patch set to be a change to either use strings like "raid1" or
> instances of the classes themselves as the sole means of specifying RAID
> levels. In short, there's no need to play with integers at all anymore
> as a means for specifying RAID levels. Calling code can be adapted as
> needed, as can the MDRaidArrayDevice class.

I have nothing against the patches nor suggested RAID level changes,
just keep in mind that Anaconda is not the only blivet user. I'd
appreciate if such API changes would be followed with some announcement
what is changed *before* release to rawhide, so I can adapt OpenLMI
accordingly.

So far there were no blivet release announces and I found that something
changed only when my tests has broken :-(. On the other side, this does
not happen often, just once or twice.

Jan


More information about the anaconda-patches mailing list