Re: Thermometer API (Re: [sysinfo] draft ready for review)

On 6 Jan 2010, at 11:42, Max Froumentin wrote:

>> What's the use case for having the 0-1 state value?  What are
>> "allowed" values, and how are they different from the minimum and
>> maximum values the system "can report"?
> 
> It's a simple way to make it easy for the webapp author to find out if the global thermal state of the system is normal, or getting near dangerous levels. The author then doesn't need to perform hardware-dependent calculations to try and average out the different temperatures reported by the system: you can't really provide a single temperature for the system, since you can have many thermometers reporting values collected at different locations in your system.

Why is "machine is in a dangerous state" the webapp author's concern, and not the runtime's?  The same comment as for the CPU API -- what's our abstraction model here? -- applies.

>> Also, are you assuming that
>> 
>> currentTemperature = min + state * (max - min)
>> 
>> (i.e., you interpolate in a linear fashion, as opposed to having,
>> say, state^2 up there)?
> 
> No, I'm not assuming any relationship between current temperatures and the general thermal state. It's left to the API implementation.

Please say that explicitly.

>> Editorial:
>> 
>> s/Celcius/Celsius/g
> 
> Fixed, thanks.

Received on Wednesday, 6 January 2010 11:17:40 UTC