[Fun] Cactus development statistics

[Vincent Massol <vm...@pivolis.com>]

Just for fun I've tried running StatCVS XML on the Cactus CVS repository.
The result is available here: http://www.apache.org/~vmassol/cactus/statcvs/

An interesting graph I think is the committer activity graph at the bottom
of http://www.apache.org/~vmassol/cactus/statcvs/authors.html.

Some analysis:
- We can see that 2003 was the best year so far, beginning of 2004 was not
that bad since end of 2004 there's less activity. 
- We can also see that Cactus committers have been participating by burst
over time (and unfortunately usually ony one burst... I'd be curious to know
why and find out if there's a way to change the trend! :-) Any idea?

Unfortunately the "software evolution graph" is unreadable. Here is what it
should look like: http://statcvs-xml.berlios.de/statcvs-xml/evolution.html

The other surprising figure for me is the LOC: 100kLOC. That's quite big for
a testing framework I think. Here's a quick breakdown that I've computed:
- Core framework: 16kLOC + 4kLOC of unit tests
- Ant integration: 23kLOC + 6kLOC of unit tests. This should be reduced
dramatically once we move to using Cargo
- Maven integration: 3kLOC + 0.5kLOC of tests
- Functional tests: 8.5kLOC
- Documentation: 30kLOC
- The rest: eclipse integration + cactus2 + misc stuff

Documentation takes quite a lot of LOC (30%+). That's good. Tests also
(total of 19kLOC = 50% of code).

The repartition code vs tests vs documentation sounds good. The only thing
we could do with is more committer activity I guess... :-)

What do you think?

Thanks
-Vincent



---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org