Matthias Andree wrote:
Rob Funk <rfunk(_at_)funknet(_dot_)net> writes:
Matthias Andree wrote:
There's no
point in saving temporary files of like 3 kByte, some file systems,
notably traditional ffs or logging ufs, are way too slow. mfs or tmpfs
or ramdisks would be ok though.
The traditional filesystems should be fast for small short-lived files
since the files will probably never even make it to disk. That's why
Unix uses asynchronous filesystems -- it's so much faster than making
sure everything goes to disk immediately.
No, sir. Linux uses async mounts by default, and has seen enough ext2fs
corruptions for this very reason. The "dirsync" patch is still not part
of the mainstream kernel. BSD has for a long time (before softdep came
into play) done anything except data writes synchronously.
I'm not sure I understand your point. I *know* Linux and BSD use
asynch mounts by default, and that's part of *my* point. Where is
your "No, sir" disagreement?
--
==============================| Guildenstern: "So there you are."
Rob Funk <rfunk(_at_)funknet(_dot_)net> | Rosencrantz: "Stark raving sane."
http://www.funknet.net/rfunk |(Tom Stoppard, Ros. & Guil. Are Dead)