Ok, this reared its head in the form of bug #1158179
I think we need to move the lock_path directory creation and removal code out of this function altogether and require the callers to do it
Using a tempdir just never makes sense
For a well known path, if this code assumes the responsibility of creating the dir, it should also remove the dir ... but the bug shows that it's not safe to do so
Ok, this reared its head in the form of bug #1158179
I think we need to move the lock_path directory creation and removal code out of this function altogether and require the callers to do it
Using a tempdir just never makes sense
For a well known path, if this code assumes the responsibility of creating the dir, it should also remove the dir ... but the bug shows that it's not safe to do so