Richard Doyle <[EMAIL PROTECTED]> writes:

> On Tue, 2007-08-07 at 09:28 -0500, Scott Balneaves wrote:
>
>> I think Linux should work that you don't need to unmount either.
>> Users NATURALLY just want to yank the stick when they're done: it's
>> logical human nature.
>
> Fair enough. Now I just need to figure out how to get high school
> students to wait until the device is done...

AFAICT, that's the entire purpose of requiring users to "eject" a
volume.  When you say "eject" (or whatever), it's a trigger to the OS
to finish flushing any pending writes, followed by a clean unmount.

With modern filesystems like ext3, as long as there are no pending
write requests the lack of a clean unmount isn't a big deal: it just
reviews the journal when it's re-inserted, and comes right back up.

If the user knows enough to care about whether there are outstanding
requests pending to the device, the user shouldn't mind "ejecting",
since that's how you can be sure that the device is really done.

If the user doesn't know enough to care (or just plain doesn't care)
about outstanding requests, there's no way to get them to wait anyway.

Sometimes i think the best way for someone to learn this is just to
get burned and lose some data because they didn't let the device wrap
up cleanly.  That may just be the BOFH in me, though.

Alternately, you could take the approach that Win95 took with floppy
disks: no caching of disk writes was allowed.  If you did an operation
which caused a write to the floppy (which was excruciatingly slow),
the entire system would hang until the write was completed.  Between
that and the grinding noise that active floppies make, you knew when
you weren't supposed pop out the disk.  

But i know i don't want to go back to that strategy.

   --dkg

Attachment: pgpCI8ctrMqkA.pgp
Description: PGP signature

-- 
edubuntu-users mailing list
[email protected]
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/edubuntu-users

Reply via email to