Is it possible to wait until the database is unlocked?

[iddm]

I tried to use r2d2 + diesel + sqlite crates for working with my sqlite database from multiple threads. However, it seems to be useless to use connection pool since everytime the database is locked the error is returned immediately without even waiting until it is unlocked by another thread. Is there a wait to specify some timeout for a connection and make it wait until one is free?

[gwenn]

In the same process, yes you can.
See sqlite3_unlock_notify.
But you will need to patch rusqlite.

[iddm]

Thanks, cool. Why nobody merges it? And why I think it should be enabled always if it is used from r2d2.

[gwenn]

See https://github.com/jgallagher/rusqlite/issues/297#issuecomment-343772632

[iddm]

Thanks

[ajensen123]

Can gwenn please provide a code example for how to use the "unlock-notify" branch. I need to write a re-entrant query function that checks the locked status of the database beforehand and then waits a random time period before retrying.

[ajensen123]

Having looked at this branch's code a little more closely, it appears that simply enabling the "unlock_notify" feature means that the internal "prepare" statement will wait until the database is unlocked before exiting. Is there any chance that it could end up waiting forever or an excessively long time?

I compiled the application with the following cargo.toml dependency defined. Is this all I have to do to enable this feature? (I'm a Rust Newbie).

rusqlite = { git = "https://github.com/gwenn/rusqlite", branch = "unlock-notify", features = ["bundled","chrono","unlock_notify"]}

[gwenn]

Currently, there is no timeout but you just have to replace the wait by a wait_timeout

[tcr]

@gwenn Can this be closed as of https://github.com/jgallagher/rusqlite/pull/349 ?

[gwenn]

Indeed, in next release.

[gwenn]

rusqlite 0.14 released