-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Simon Slavin wrote: > Perhaps this passage could be rephrased to warn explicitly about NFS > rather than about the more general "files on a network filesystem".
As a general rule network filesystems are buggy. Local filesystems get to make all the decisions themselves - there is no other party. With remote filesystems everything is passed to the remote server which makes all the decisions. This of course is eye wateringly slow adding latency to every filesystem operation. So the network clients occasionally make a decision locally instead of sending it to the server. (This is also a *lot* easier to code.) Earlier NFS releases were remarkably lax on the client side - the Unix Hater's Guide even has an entire entertaining chapter on it. SQLite exercises codepaths that aren't particularly normal compared to most applications and locking is even rarer. Unless you can guarantee *all* client side code, the server side and interactions with multiple clients is correct then there is the possibility of corrupting SQLite files. Based on past experience there is also the probability they will be corrupted. Are you willing to stake your reputation and whatever else on there being bug free implementations of AFP and SMB. (BTW in a past life I coded an SMB server - the other clients and servers out there are definitely not bug free :-) Users of SQLite won't appreciate their databases being just a little bit corrupted infrequently. Roger -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkrVOV4ACgkQmOOfHg372QTxkgCfVrY2bpmoDtfw2rI2pnsG0o8G uRkAoIRFY8A1sKZRFTyV1/2iqcxH4a6G =jv8p -----END PGP SIGNATURE----- _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
