This repo contains tools to analyze the inclusions of a given set of files in the codebase over time. There are two fundamental operations:

• Generating a database that has information about the inclusions of a set of files.
• Generating analyses that process the information.

We explicitly separate these steps because the first is slow and usually done only once for a given Chromium revision and set of files to be analyzed, while the second is fast and may be done many times for a given database.

TO GENERATE A SIGNIN DATABASE

Check out Chromium at the desired revision (most likely at HEAD). Run the following command: ./generate_inclusions.py /path/to/chromium/src data/database_configs/signin_internals.py data/databases/

TO GENERATE SIGNIN ANALYSES

The easiest thing to do is just regenerate all the analyses, which is still fast and shouldn't change anything besides adding data for any new databases and/or analysis configs that you've added.

./scripts/generate_signin_analyses.sh

Once the script is done, the result will be in data/databases/signin_internals/<git_hash>/analyses/

TO ADD THE RESULTS OF AN ANALYSIS TO THE SPREADSHEETS

The spreadsheets live here:

https://docs.google.com/spreadsheets/d/1KJmaPq5vcDuoeADY0bkQKlNhL73peslwV9QXXmkpEMo/edit#gid=353755218 (request access if you think you need it)

To add the result of a new analysis to the spreadsheets, do the following for each analysis concerned:

• Run scripts/collate_progress_over_time.sh
• Any new data should be at the bottom. Copy it.
• Paste it into the relevant sheet on the spreadsheet, click the paste icon and click "Split text to columns".
• Pull down on the lowest filled-in cell in the "total" column so that the total gets calculated for the new data.
• Observe that the chart gets populated with the new data.

TO MAKE CHANGES TO THE CODEBASE

After making changes, run ./test/run_tests.py

If feasible, add/update any tests if making non-trivial changes.

TODO:

• Find a better place to store data than in the repo. It's unclear where else we would store it to allow (a) sharing across users and (b) easy analysis by in-repo tooling.