-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Simon Slavin wrote: > What I will say is this: I see no reason why there > should be more bugs in the AFP locking code than there are in the > locking code that comes into play if you're accessing a file on your > own hard disk.
In which case you missed my entire point :-) When doing local code nothing can change under your feet since inode objects are shared. When there is a remote system implementing a call that requires several checks of the remote end then code often uses cached information or takes shortcuts. When the local API exactly matches a protocol request exactly then it isn't a problem but in various cases it won't. Quite simply getting a networked filesystem to be "correct" is significantly harder than a local filesystem. Get locking correct is even harder. Having reasonable performance while doing all this is even harder still. > I expect Apple tries to fix bugs in its AFP code pretty quickly. Did you notice that Apple implemented their own locking mechanisms for SQLite because of mismatches between Apple APIs and filesystems? And that you will get corruption unless *all* access to a database is using exactly the same locking mechanism. This is easier to guarantee with local processes but a lot harder when there are remote ones. > I expect Microsoft pays just as much attention to bugs in its SMB > code. The bugs the vendors care about are backwards compatibility and performance. Show a program that used to work and doesn't now and they'll pay attention. But other than that they don't care since they don't have to. Roger -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkrVY3UACgkQmOOfHg372QR7cACcC9b67MmCZOp/kPE/6h/+lXsY H0sAoIFKjR8Q6rF/u6rTcbqD04UzhRVN =kD3t -----END PGP SIGNATURE----- _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
