|
@@ -22,6 +22,8 @@ to get the annual burnout plot, set granularity to 365. Sampling is the
|
|
|
frequency with which the burnout is snapshotted. The smaller the value,
|
|
|
the more smooth is the plot but the more work is done.
|
|
|
|
|
|
+
|
|
|
+
|
|
|
###Installation
|
|
|
You are going to need Go and Python 2 or 3.
|
|
|
```
|
|
@@ -36,7 +38,7 @@ wget https://github.com/src-d/hercules/raw/master/labours.py
|
|
|
hercules https://github.com/src-d/go-git | python3 labours.py
|
|
|
# Use "file system" go-git backend and print the raw data.
|
|
|
hercules /path/to/cloned/go-git
|
|
|
-# Use "file system" go-git backend, cache the cloned repository to /tmp/repo-cache and display the plot.
|
|
|
+# Use "file system" go-git backend, cache the cloned repository to /tmp/repo-cache and display the plot.
|
|
|
hercules https://github.com/git/git /tmp/repo-cache | python3 labours.py
|
|
|
|
|
|
# Now something fun
|