[Teammetrics-discuss] NNTPStat completed successfully.

Andreas Tille andreas at an3as.eu
Sat Nov 5 00:00:11 UTC 2011


On Fri, Nov 04, 2011 at 10:49:55PM +0530, Sukhbir Singh wrote:
> The year is 1989 because that is what the year was in the message itself!

Well, the header information of the mail says 1989 - I probably could
fake any date by simple resetting my computers clock.
 
> So our code is OK but this header is not. I can add a check for this
> also (if year < 1990 or something) but then there is only one such
> message so doesn't make sense.

The problem is, that we *see* the problem in this message but I bet that
there are more such problems we can not see.

> If you feel otherwise, please let me
> know and I will fix it.

My old algorithm was not reading the date field but rather was obtaining
year & month information from the month of the archive in question.  IMHO
this is more safe and prevents surprises like this one.
 
> > In Debian Jr list I guess 'Andreas T' and 'Tille A' will be the
> > same person. :-)
> 
> We had 'TilleA' but not 'Tille A' :) Fixed and added both.

Well, I guess this single 'A' in the end is because of stripping the
(assumed) last name - the database has no string 'Tille A' and thus this
is surely not the solution. I checked the database via:

SELECT name, count(*) from listarchives where name ilike '%tille%' group by name;

and found the following variants of my name:

Tille, Andreas
<tille at debian.org
'Andreas Tille'
<tillea at rki.de

(BTW, Alexander Tille is my son :-)).

I somehow have the impression that updatenames.py fails in some
circumstances.
 
> > The Debian Lex graph looks *very* different from what I just calculated
> > with the old code
> >
> >    http://blends.debian.net/liststats/authorstat_lex.png
> 
>  archive_date
> --------------
>  2007-12-14
>  2007-12-14
> 
> The mbox archive we create from Gmane for `debian-lex` starts at 2007
> instead of the 2003 in your code and that is why there is the
> difference... we can't do anything in this case unless we get the mbox
> archives (frankly, which I don't think we will, you have already asked
> patiently lots and lots of times).

I will try again soonish - but I told you it will take some time.

However, I'm not convinced that the Gmane workaround was a good idea in
the first place and perhaps it would have been better to follow my
original idea to parse the web archive.
 
> I have already signed up as a mentor! Yayay! I have included some
> project for Team Metrics also.

Cool! :-)
 
> Yup, that's right. I fear that we might not get many applications for
> Team Metrics also, but oh well, we will see and I will try to spread
> the word in IRC also if some students come :)

Good luck!
 
> PS: When you are free, please give `commitstat` a run also.

I did as well and it was running without uncovering any problem.
I need to check the data obtained.  Just forget to mention this.

Kind regards

        Andreas.

-- 
http://fam-tille.de



More information about the Teammetrics-discuss mailing list