--On 2004-1-30 1:20 AM +0530 Madan Ganesh Velayudham
<mganesh(_at_)india(_dot_)hp(_dot_)com> wrote:
Please share views on the following :
"SIEVE Statistics", Madan Velayudham, 03-Dec-03. (9179 bytes)
Some comments based on a quick scan of the document ...
Section 4:
* "action" needs to be more clearly defined
Section 5:
* Why two capabilities? Both commands are required, so there's no
point in advertising two capability strings. The capability should be
changed to something like "STATISTICS".
Section 6:
* What does a "zero valued action counter" mean? Surely '0' is a valid
counter value, meaning the action has never been triggered. I think
what you're implying here is that a '0' value means the counter for the
action isn't implemented. If that's the case, just don't return a
response for that action.
* Presumably GETSTAT is only valid in authenticated state? (Section 7
explicitly calls this out, so section 6 should as well.)
Section 8:
* The grammar (and semantic description) for the command responses is
missing.
Section 9:
* Since there's no way (that I can see) to access another users counter
data (through the protocol), this text doesn't seem relevant.
General:
* How useful is the returned data? The fact that an action took place
doesn't really tell me anything. I would be much more interested in
seeing how often each rule fired, as that would allow me to optimize my
rulesets.
--lyndon