Re: [HACKYSTAT-DEV-L] 'demo' telemetry streams/charts/reports

[(Cedric) Qin ZHANG]

(1) Approach 1:
Only admin can have global telemetry definitions. I assume no one wants
to make "testtelemetryuser" admin. I can change telemetry implementation
so that everyone can define global definitions, but is that a good thing?
(2) Approach 2:
Reducers come from different modules. Different configurations may have
different set of  (or even 0) reducers available. How can we guarantee
that the downloadable definition file always uses reducers available? Do
we assume that people always download Hackystat-UH configuration, or a
superset of it?
Cheers,
Cedric
Philip Johnson wrote:
Cam mentioned to me this morning that he found the telemetry reports and
charts in the
documentation to be pretty cool but was surprised and unsure of what to
do next when they
did not appear in his downloaded version of the system. (The problem is
that the
telemetry definitions are on the public server in my account as part of
my 'global'
definitions, but are not included in the stable/daily builds.)
I've come up with two approaches on how to address this.
Approach One: We create a set of "demo" streams, charts, and reports
that are included in
a 'testtelemetryuser' account that would be installed into any
configuration containing
hackyTelemetry.  These would have 'global' access and thus be available
in all accounts.
The name of each stream/chart/report would start with 'Demo' to clarify
where it came
from.
That way new users could immediately start using a really basic set of
telemetry right
away without having to make any definitions themselves.
We could also write unit tests against these to ensure that they work
correctly as we
modify the language/environment/reduction functions.
Approach Two: Provide a downloadable user.doc.telemetry.xml file that
the admin can put
into a user directory, restart the server, and get a set of
shrink-wrapped 'standard'
telemetry definitions.
No ability to write unit tests against them, but we could avoid the
'Demo' tag.
Any thoughts from anyone about which approach is better? Does anyone
have an Approach
Three?
Cheers,
Philip