[Teammetrics-discuss] Linking Metrics Together

[Sukhbir Singh]

Hi Vipin,

> As I mentioned in an earlier mail, we could link different metrics of
> the same project together. This in my opinion will give a better
> picture of a teams functioning.
>
> This can be done in two ways, either at the database level or at the
> code level.
>
> 1) Database Level
> [...]
>
> 2) Code level
> [...]

> These are two methods that came to my mind for linking metrics
> together. Which one is preferred? If it is possible in some other way,
> please let me know.

I personally feel that modifying the database is an overkill for this.
For such situations (at least in Team Metrics), I have always used a
config file. In the Team Metrics code, see
`etc/teammetrics/commitinfo.conf'. This config files contains the
different teams under their appropriate 'type'. Similarly, there is
'listinfo.conf'.

So, I prefer the Python code if you ask me. We just load the config
file and get our job done.

A sample config file for this can look like:

    [teammetrics]
    list = teammetrics-discuss
    repository = teammetrics

    [soc-coordination]
    list = soc-coordination

What I suggest is that you do this: say the data for teammetrics
mailing list is requested ... you fetch the teammetrics section from
the config file, read the 'list' value and fetch
'teammetrics-discuss'.

I guess this is what you were aiming to do, right? As far as
populating this config file is concerned, I can help you with that.

> Regarding the data, can we crowdsource it? Say request the community
> to fill up a form with project name, mailing list id and all other
> relevant information?

I had thought of this once. For (most) teams, we already have this
ready. For others, we can request them.

> I have access to only tables in the database. They are 'commitstat'
> and 'listarchives'. Where are the other metrics (bugs, uploadstats)?
> If there are some other relevant tables, I kindly request access to
> them :)

Andreas, your department :) I think there is no database and the
graphs are generated on the fly by reading data from the Ultimate
Debian Database [0].

Feel free to ask me if anything is not clear.

[0] - http://udd.debian.org/

-- 
Sukhbir