[rrd-developers] Re: Hallo!

[Tobias Weingartner]

On Tuesday, July 27, Tobias Oetiker wrote:
> 
> uff ... great ... 

Again, thank you.  Now, two suggestions.  1 is easy to code (from the quick
peek I had), the other likely a little harder.

1) Along with the current IF,GT,etc RPN notation, a DUP would be exceedingly
usefull.  Especially within cricket, where the data-source is implied in the
scale= portion.  Currently I could use that to remove "spikes" in the graph.
(My current RPN is: 'CDEF:sds1=ds1,8,*,ds1,8,*,10000000,LT,*', but within
cricket, all I can do is something like: 'scale=8,*', which is the tail end
of the CDEF.  If I had a DUP function, I could rewrite the above LT stuff...)

2) One more consolidation function.  However, this function could take an
argument to it, so I'm not sure how it would be invoked/etc.  The function
is something like 'WEIGHTED', which would be a weighted average, using some
sort of exponential, linear, or other function, to weigh the individual points
within the average.  One possibility, would be to have the ability to store
the weight of each point within the RRD file.  That way the "function" could
be hand crafted whichever way you wish it to be.  Common things like an expo-
nential, or linearly weighted function could be "built-in".

Something along the lines of:

RRA:WEIGHTED(linear,0):0.1:12:300		# Equivelant to 'AVERAGE'
RRA:WEIGHTED(linear,-1):0.1:12:300
RRA:WEIGHTED(exp,0.5):0.1:12:300
RRA:WEIGHTED(0.5,0.2,0.1,0.05,0.05,0.05,0.05):0.1:7:300

WEIGHTED(linear,X) -> X == slope


I'm not sure if I'm making much sense here...

--Toby.

--
* To unsubscribe from the rrd-developers mailing list, send a message with the
  subject: unsubscribe to rrd-developers-request at list.ee.ethz.ch