added some remarks on CDEF versus VDEF to the docs -- Pablo Sanchez

[rrdtool.git] / doc / rrdgraph_data.pod

diff --git a/doc/rrdgraph_data.pod b/doc/rrdgraph_data.pod
index 95b1ec7..02f9663 100644 (file)
--- a/doc/rrdgraph_data.pod
+++ b/doc/rrdgraph_data.pod
@@ -90,6 +90,34 @@ the script, just as if it were generated by a B<DEF> instruction.
 
 Example: C<CDEF:mydatabits=mydata,8,*>
 
+=head1 About CDEF versus VDEF
+
+At some point in processing, B<RRDtool> has gathered an array of rates
+ready to display.
+
+B<CDEF> works on such an array.  For example, I<CDEF:new=ds0,8,*>
+would multiply each of the array members by eight (probably
+transforming bytes into bits). The result is an array containing the
+new values.
+
+B<VDEF> also works on such an array but in a different way.  For
+example, I<VDEF:max=ds0,MAXIMUM> would scan each of the array members
+and store the maximum value.
+
+=head2 When do you use B<VDEF> versus B<CDEF>?  
+
+Use B<CDEF> to transform your data prior to graphing.  In the above
+example, we'd use a B<CDEF> to transform bytes to bits before
+graphing the bits.
+
+You use a B<VDEF> if you want I<max(1,5,3,2,4)> to return five which
+would be displayed in the graph's legend (to answer, what was the
+maximum value during the graph period).
+
+If you want to apply 'complex' operations to the result of a VDEF you have
+to use a CDEF again since VDEFs only look like RPN expressions, they aren't
+realy.
+
 =head1 SEE ALSO
 
 L<rrdgraph> gives an overview of how B<rrdtool graph> works.