Unstable Branch Testing

[HerbFargus]

Just logging an official ticket to discuss the statistical advantages of the SQLite database over the current xml database, particularly as it pertains to large romsets and performance.

[Shmoopty]

This issue pertains to both stability and performance. It's unusual that quitting or restarting any program could take several minutes. But regardless of romset size it's also unusual that, hours after scraping has occurred or playcounts have been altered or any metadata has been refined, a crash or pulled power plug or "kill -9" or "sudo reboot" would undo all of those changes as if they had never happened.

[joolswills]

Worth keeping an eye on this fork - https://github.com/Herdinger/EmulationStation - if it continues to be developed, and gets decent improvements etc. There is a PR open with a bunch of stuff on top of the sqlite3 version, but needs fixing/rebasing against the branch it's supposed to be merged to.

[MisterTea]

I created a PR with sqlite support but the merge was incomplete (didn't fully support directlaunch). #31

One of the reasons not to continue with the merge was that "issues with the sqlite branch" was found (don't know what those issues were). I've been running with the sqlite branch for several months now without issues. I've enjoyed immediate updating of playcounts and startup/shutdown time of milliseconds instead of minutes.