[metrics-team] Adding churn visualization to Metrics

[Philipp Winter]

Hi Karsten,

On Tue, Jan 26, 2016 at 10:57:44AM +0100, Karsten Loesing wrote:
> Do you want to take a look at this guide and give me some feedback
> what parts need more explanation?

Of course, thanks for writing this!  That should be enough to get me
started.  I added some minor feedback inline.

>  - What Metrics is not: The typical Tor researcher or Tor developer
> would probably want to dive deeper into the data to learn even more
> about the Tor network.  But in contrast to the average Metrics user
> they could just fetch the original data from CollecTor and run their
> own analysis.  Metrics is not trying to be the solution for everyone.
> If we have to choose, we're aiming for simplicity than for
                                                    ^^^^^^^^
I would write "instead of".

>  - Data-processing modules: The bulk of Metrics code is running in the
> background, invisible to Metrics users.  It's the code that takes
> CollecTor data as input and that produces .csv files that are the
> basis for graphs and tables on Metrics.  There's usually one such
> module per generated .csv file that focuses on a different aspect of
> Tor network data.  All these modules are periodically executed by the
> system's cron daemon, independent of user requests to the website part
> of Metrics.  See the modules/ subdirectory for the existing
> data-processing modules.  Note that modules don't have to be written
> in Java even though that's currently the case for all of them.  The
> only requirement is that there's a shell script to run the module
> using packages available in Debian stable.  The remaining components
> of Metrics are all related to its website part.

I would mention explicitly that you are talking about the metrics-web
git repository, and maybe even provide the URL:
<https://gitweb.torproject.org/metrics-web.git/>

Cheers,
Philipp