Collecting trends retroactively

[martinrleon]

Thank you for the git-fame tool, it’s a great resource for me.

I want to develop a dashboard of metrics-over-time using your tool. What is the best approach for collecting ‘past snapshots’ - meaning, if I want to see the trends in metrics for the past year, month over month, what is the best approach?

I started using ‘--since’ and ‘--until’ to kind of partition the results as if they were being collected over time, one month at a time, starting from a ‘shallow’ clone going back to include only 1 year of commits. So it would be '--since 2025-07-01’ '--until 2025-07-31’, then the same for 2025-06, 2025-05, 2025-04, and so on until 2026-06 to catch me up to the current month.

The collected data monthly snapshots (repo totals / contributor totals in json) would be captured in a DB and then graphs would show trends in commits, files, hours, etc, either at the individual level or at the repo level month-by-month.

My biggest concern is to accurately capture and represent these trends - does the above approach look reasonable / correct?

Any suggestions appreciated.

[casperdcl]

might be somewhat related:

1. https://github.com/casperdcl/git-fame/graphs/contributors?from=20%2F07%2F2024
2. git punchcard:

git punchcard -p month/wday -- --since=1.year --author="Casper"

[martinrleon]

I left out a few important goals. I want to be able to do chart comparisons across repos / repo orgs (multiple repos). I would also want to be able graph individual trends and to distinguish work done by contractors and 'bot' committers (I would need a table that maps emails to people that allows me to categorizes them). And finally, I would want to be able to show what kinds of code is being compared, and to be able to exclude non-code files / folders from the collected data.

git-punchcard seems to run very quickly but it only has one representation of the data and I wouldn't be able to compare committer activity or that of one repo / repo org against another repo / repo org on the same graph without the detailed data.

I have some POC charts to demonstrate what the end goal is.

With the detailed data from git-fame I feel this is achievable but want to make sure I'm correct in this assessment / approach. Or if there is a better existing tool to get there more efficiently, I would be open to that as well.