The tree model currently merges the data by IP, and not by function name. The latter would be more useful, I think. It would esp. allow the model to be reused as-is for the flame graph visualization. The issue is finding an efficient storage and visualization that still allows to show per-line cost.
some ideas on how to do it:
- ask user on how he wants to group the data, configure columns accordingly and potentially re-read the data file to apply the aggregation
- always group by function, add annotate feature that shows file:line costs