Re: [PATCH 2/5] Xl interface change plus changes to code it impacts

[Kamala Narasimhan <kamala.narasimhan@gmail.com>]

Ian Campbell wrote:
> On Tue, 2011-02-08 at 18:42 +0000, Kamala Narasimhan wrote:
>>>> Did we decide to leave DISK_BACKEND_DEFAULT (i.e. libxl chooses based on
>>>> the image type, host's backend support etc) for 4.2? I don't mind if we
>>>> did but does that make "block-dev-type" as described in patch 1/5
>>>> non-optional? (and therefore not really deprecated)
>>> Maybe this is handled in xl by patch 3/5?
>>>
>>> (I should really apply the patch and read the result instead of trying
>>> to decode the meaning from the patch form)
>>>
>> Yes, a fallback value is set in 3/5 but patch 5/5 will remove that fallback and
>> do appropriate validation to better handle this.
> 
> Patch 3/5 is all on the xl side though, right? I was meaning a default
> "do the right thing" at the libxl interface level, which given time
> might become the only option...
> 
> I didn't see a patch 5/5 what is in it?
> 
Based on further discussion we decided to go with patch 1 & 2 and a couple of
minor patches on top for 4.1 and apply the rest for 4.2.  I will send patch 5
along with 3 & 4 post 4.1.

Patch 5 is all validation code which will be pretty much in libxl and it will
take care of the fallback,  For now we do nothing more than what we already do IMO.

Kamala