Re: [PATCH 02/10] numa: introduce MachineClass::forbid_asymmetrical_numa

[John Snow]

On 8/20/20 12:51 PM, Eduardo Habkost wrote:
> In either case, it sounds like this won't be a common constraint
> and I now agree with your original suggestion of doing this in
> machine initialization code.

I'm late to this chat, but not every constraint needs to be modeled as 
data -- it's okay to do validation in the machine initialization code.

(It can always be changed later!)

I have been more concerned with schema-based validation of the shape of 
data, but which boards support NUMA or not falls perfectly into the 
realm of semantic validation that is outside the realm of static data 
validation, I think.

--js