Just to remind people, I'm currently writing a visualization that looks
at the relationship between code & communication. Since I have not
been able to get an archived version of this mailing list, and I dont
have time right now for screen scrapping off source forge, I ran my
visualization against statcvs but without the communication. You can
find a copy of it at
http://bonobo.media.mit.edu:8080/~aaron/statcvs.pdf There are still
seems to be a rendering error (where people will cross over each
other), and this pdf is not a screen shot thus loosing the interactive
exploration of my program, as well as dates. But I'm curious to get a
reaction from a similar project (w/o the communication). I created
this by looking at each file revision and who owns the code over time.
Each color represents a person, and the height is proportional to the
total amount of code they own as time increases (starting from the
beginning of the CVS archive). There is some quantization from nearby
commits to create smoother lines depending on the scaling of the
x-axis.... but since this is a 2 year few the quantization is decently
high. Also this does not include some deleted files that CVS couldn't
access (gtk-- doesnt have this error but statcvs does). To combat the
visual artifact of one person's commit looking like everyone committed
(since the whole stack rises), you'll see some white lines going over
the increases... this indicates who committed and is responsible
(possible for multiple). The people are in order of color lukasz,
cyganiak, manschu, jentzsch, farkas, squig.
Is this what you'd expected statcvs? Does the data look inaccurate,
and if not, does it reflect what how you thought about the project in
comparison to statcvs's outputs of itself?
Thanks for any feedback,
Aaron Zinman az...@me...
MIT Media Lab E15-390
Sociable Media Group 617.452.5606
|
