This document summarizes a presentation about visualizing project data and trends using graphs generated from code repository commit data. It discusses analyzing metrics like lines of code over time, which developers work together, which areas of code see the most changes, and how many people typically work on stories. The presentation demonstrates parsing git commit data and experimenting with different graphing libraries to visualize trends in the codebase evolution and team collaboration. It aims to provide insights on productivity and opportunities for improvement.
2. Curious to see the
trend with lines of
code
How big are our
commits?
Do people work in
the same area of
the code?
Do the same
people pair with
each other?
How many people
work on an average
story?
Why?
3. “Budget for failure by designing and building
toy systems that are similar to the systems
you build at work in toolset but not in scope.”
Breakable Toys
We’re using a library called Step here to pass on the commits to the next function where it gets sent to the callback after being parsed
Finally got to the point where it made sense to write unit tests around part of the code. Before that it was much quicker to just refresh and have a look what was going on
Build a word cloud instead
Build a word cloud instead
Build a word cloud instead
Build a word cloud instead
Build a word cloud instead
Build a word cloud instead
Did it all on the fly to start with but Then realised it was way too slowSo pushed it into mongo
If it’s actually synchronous – indicated by Nesting loads of functions then maybe It’s not a job for node
If it’s actually synchronous – indicated by Nesting loads of functions then maybe It’s not a job for node
If it’s actually synchronous – indicated by Nesting loads of functions then maybe It’s not a job for node
If it’s actually synchronous – indicated by Nesting loads of functions then maybe It’s not a job for node
If it’s actually synchronous – indicated by Nesting loads of functions then maybe It’s not a job for node
If it’s actually synchronous – indicated by Nesting loads of functions then maybe It’s not a job for node
If it’s actually synchronous – indicated by Nesting loads of functions then maybe It’s not a job for node
High charts seems to be the best general purpose library but protovis is better if you want to make custom charts. Protovis is now d3.js
You could go on holiday and then use these graphs to help you work out what had happened in your absence
It was suggested that by making these graphs available that we’d encourage people to game the system to try and get to the top of the charts. In this case we weren’t sure if that’s necessarily bad because the Behaviour we wanted to encourage was people checking in in smaller increments than they had been previously.
It was suggested that by making these graphs available that we’d encourage people to game the system to try and get to the top of the charts. In this case we weren’t sure if that’s necessarily bad because the Behaviour we wanted to encourage was people checking in in smaller increments than they had been previously.