>>>
>>> There is no such clash in VirtioPMemDeviceInfo.  You could name the
>>> member "addr" there.  But that would trade the inconsistency with
>>> PCDIMMDeviceInfo for an inconsistency with the device property.
>>>
>>> Is this correct?
>>>
>>
>> Yes, I chose to name it like the property. (that felt to be the right
>> thing). As far as I see this is perfectly fine.
> 
> Now work that into the commit message or perhaps a comment, please :)

Yes, once this get's picked up in the !RFC version of virtio-pmem, we
should highlight that (and best, split VirtioPMemDeviceInfo introduction
off from the remaining part).

> 
>>                                                 It's unfortunate but we
>> can't do anything about it.
> 
> Well, if we really, really wanted to, we could: rename pc-dimm's
> property, keep the old name as a deprecated alias.  Would that be better
> than the inconsistency, and the code you need to work around it?  You
> decide.
> 

I had the same in mind initially but considered it not useful right now.
There is not that much code to work around that (it's basically just two
functions get_addr() and set_addr() on the memory device interface), and
this way we can at least keep property setting/getting code out of
memory-device code (I don't consider properties useful when it comes to
internal interfaces).

If we ever see a problem with that, we can introduce the alias (+
optionally deprecate) later.

Thanks!

-- 

Thanks,

David / dhildenb

Reply via email to