[Teammetrics-discuss] Changes to GitStat.

Andreas Tille andreas at an3as.eu
Sun Aug 21 19:38:00 UTC 2011


On Sun, Aug 21, 2011 at 07:00:23PM +0530, Sukhbir Singh wrote:
> |      package
> ------------------------------------------+---------------+------------+-------------------
>  761c0772e695db1199e5116d6ba06535cab0c993 | Shaun Jackman | debian-med | igv
>  761c0772e695db1199e5116d6ba06535cab0c993 | Shaun Jackman | debian-med
> | igv-manuel
>  761c0772e695db1199e5116d6ba06535cab0c993 | Shaun Jackman | debian-med
> | igv-to_be_removed
> 
> So why are there more than one packages with the same hash? :) I don't know.

Ahhhh, I remember this case.  The igv repository was broken at some point in
time and people seem to have decided to keep copies of it.  So considering this
IMHO the proper behaviour is:

   try:
	insert ...
   except <duplicate key>
	issue warnig that in repository XY Id was duplicated
        continue and simply ignore this commit
 
> So two things can be done:
> 
> 1. Either we set the primary key on the package column (like I have done).
> 2. We set it on the project column only (blends.d.n and our
> repository) and discard multiple commits of the same hash.

Nothing like this.  The duplicated commits should be ignored - that is not
new information, just copies.
 
> I did not go for (2) by myself because of multiple packages having the
> same hash.

IMHO the same hash is a sign for duplicated code and should not be
counted twice.
 
Kind regards

     Andreas.

-- 
http://fam-tille.de



More information about the Teammetrics-discuss mailing list