Closed cakirke closed 8 years ago
I suppose stealing "stale" lockfiles is pretty common, so I'm ok with this. But it makes me a bit nervous. Personally, I would only want to steal a lockfile if I knew the current holder was dead and/or I gave permission to steal it (it seems that setting the stale timeout is tantamount to giving permission). Would it be prudent to issue a warning when this happens?
Note: SQLite has its own locks, so that gives us a bit of a safety net.
+1 on nervous, agree on setting environment = acknowledgement/permission but an appropriate message wouldn't hurt
Hmm, I thought this was included in the 0.09998 release, but it appears that I never merged it.
address #201 - support additional File::NFSLock capabilities, stale lock expiry, errstr in thrown exceptions - will include lock holders in next version of File::NFSLock