[noduerme]

It's really only the write / ACID features of WebSQL that are interesting to me, though. There are lots of ways to get a high performance read on key/value pairs if you've already downloaded all the data. We don't need a new methodology for that. What's interesting is the ability for web app storage to respect foreign key constraints and transactions within a local data model, without having to rely on rolling your own middleware to resolve inconsistencies.