Simon (and others), Many thanks for your responses. This clears matters up a good bit.
It just means that I will have to rethink how I implement this database 'delete' operation. Cheers, Dennis ---------- Forwarded message ---------- From: Dennis Suehr <den...@suehr.me.uk> Date: 12 November 2010 16:44 Subject: Re: [sqlite] Database deleted from file system while other open connections exist. To: slav...@bigfraud.org Cc: sqlite-users <sqlite-users@sqlite.org> Simon, Many thanks for your prompt and thorough response. Just a couple of follow-up questions if I may: Firstly, a bit more background information. The system will be running on Linux (kernel > 2.6.0) and all accesses to the database will be via local disk (ext3 or similar). 1) I appreciate your explanation about any any open connections to the file still being able to write to it until they close it. I suppose that 'fuser' could be used to send a SIGTERM or similar to the processes holding open file handles. I will have to give some additional thought to how I go about this 'delete' operation. 2) Regarding, SQLite and open handles to a database, that is fair enough that it does not have a mechanism send alerts to all processes connected to a given DB. However, I am surprised that it does not at least possess a list of all open handles to a given database. If I had that, then I could close all DB connections either before the delete or after. Are you sure that such a list does not exist? Wouldn't it use it for mechanisms like 'unlock notify' and 'busy handler', etc? 3) Finally, am I correct in thinking that if I used 'pragma locking exclusive' that I can lock the entire database even after closing my connection? Cheers, Dennis _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
