On 18 Nov 2009, at 8:38pm, priimak wrote: > I understand that. However, that would not present a problem if I access > database by its symlink for reading only and database directly if I am > updating isn't?
According to the documentation that could be a problem. The update application creates a journal file in the directory where the database file actually is. The application which reads the database file looks for a journal in the directory where the alias is, and doesn't find it. Please note that I haven't read the source code for sqlite3, and I don't know if that's really how it works. > By the way the reason for that usage ( though there are work arounds ) > is that I keep two instances of db file a.db.1 and a.db.2 with symlink > a.db pointing to one or there db file. If a.db -> a.db.1 I apply update > to a.db.2, swap symlink and then update a.db.2. Symlink a.db used by > webapp, which performs *only* select queries. Rather than use a symbolic link, store something in the database files themselves which says which one is being updated. Or use two different tables in the same database file. I understand your use of symbolic links and it makes sense with atomic files -- files where you open the file, read the whole thing, then close it. But it would seem to be a problem with SQLite files. Simon. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users