Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo

Close

#72 Improve Commit activity scatter chart

open
nobody
5
2007-03-25
2007-03-25
Benoit Xhenseval
No

Hi

I would suggest to modify very slightly the activity chart to show new or update by using a different colour for the dot.

e.g.
http://www.statsvn.org/demo/qalab/commitscatterauthors.png

Blue could be new and green amend?

Let me know if you'd like me to have a go and provide a patch.

Regards

Benoit

Discussion

  • Logged In: YES
    user_id=584620
    Originator: NO

    You mean, use a different color if a new file was committed?

    Practical problem: A commit can add new files *and* modify existing files, how to color it?

    General problem: This change might add more noise than information. What question does this help answering? What “aha!” effect might it cause? Let's not add things just because we can.

     
  • Logged In: YES
    user_id=358115
    Originator: YES

    My idea was to show if a developer was primarily creating new files at some point and then at some other time, primarily modifying it.
    It is a bit like the other activity report, so dots could be green/yellow (or whatever).

    B

     
  • Logged In: YES
    user_id=584620
    Originator: NO

    Benoit, I agree that showing if someone adds new stuff or modifies existing stuff, and how it changes over time, is interesting and a good idea.

    The suggested chart doesn't do this too well. The dots are all the same size, so if someone does 10 commits, 9 of them changing a single LOC, and 1 adding five hundred new files, it would appear as nine “modify” dots and one “add” dot, which misrepresents what was going on. Unrelated information (how the developer chooses to break up his work into individual commits) dominates the actual data (how much was added or changed).

    We could try playing with the size of the dots (larger for big commits), and maybe have a gradient of colors instead of just two colors, but I'm afraid this wouldn't work well for very active authors on old projects.

    I'd say that number of lines added/modified, like measured for the other activity report, is a more representative measure than number of files added/modified.