Current setup
- Runs queries like collection:foo+language:bhangra (one per project I think)
- Does this in Jake's python tool
- Result is big directory of files 10-400TB
- Sneaker net directory
- Either
- Runs on server in India with little internet access
- Gives hard disk to viewer
Cares about
- Update since 400TB would be quicker if can check (done)
- Run on local disk, not just server - hard dweb-mirror issue#63
- Compatability with Jake's tool (need to discuss) dweb-mirror issue#64
- Possibly Branding if becomes part of bigger project (need to discuss policy)