[Libmetrics-devel] Re: Metrics 2.0?

[Luke K. <lu...@ma...>]

On Mon, 2005-04-11 at 11:17 +1000, Tim Nelson wrote:
> On Fri, 8 Apr 2005, Luke Kanies wrote:
> > Well, the real problem here is that the functions aren't meant to take
> > arguments, and the disk function obviously needs to do so.
> 
>  	I was kinda wanting to myself for something else, but if we use 
> the "get function" thing I talk about in the other mail, that'll solve the 
> problem for me.  But not for disk :).  And disks aren't the only things 
> which have multiples.  Eg.
> -	CPUs
> -	Load averages
> -	Network interfaces
> 
>  	As time goes on, I suspect that multiples of things are going to 
> become more popular.

I agree with that statement, but I don't think your solution quite
covers it, unless the va_start stuff does the rest.  It seems that for
most things that involve multiple numbers (disks, cpu, load, network),
each type of metric needs a function to collect all of the information
and store it in some globally accessable place, and then there needs to
be a specifically named metric for each instance (each disk, each cpu
type, etc.) that one can use to retrieve each of the values.

Is that basically what you had in mind?  Is the collection function what
you are calling 'va_start'?

Do you have an idea for such an architecture?

BTW, I've CC'ed the libmetrics-devel list, so we can start getting these
emails on record.  Please continue in that vein.

-- 
I also realize I can no longer spell "pseudo". My finger have now
hard-wired "sudo". Thanks, Unix! I also cannot uppercase "cat" for
the life of me. --Jonathan Rentzsch
---------------------------------------------------------------------
Luke Kanies | http://reductivelabs.com | http://config.sage.org