- Switch over to using a SQL database for the build summaries rather than text files. This should allow easier searching and is required for some of the functionality listed below.
- This requires moving away from sqlite which only allows one concurrent writer. The code supports this but we need to change the setup on fn.
- Only look at oldrevs/ directory, never at upload/. Instead, import-and-analyse.py should be polling for updates continuously. This means the web interface never has to parse more than one build log file for a request.
- Should be trivial once we move to a SQL database.
- Add revision table for each project including links to the builds that have happened for each revision. This requires the sql database or keeping the .status files around for longer.
- Use python subunit parser rather than nasty regular expressions.
- Ability to edit hosts from the web UI
- Ability to check up on individual tests across all hosts
- For a single host, ability to see the full code diff between two historic builds.