> In discussing this problem, several people had possible
> solutions which were
> all shot down with a similar counterargument: the kernel
> can't assume that
> the user who is using the floppy drive is at the console.
>
> It is not 100% guaranteed that the user is at the console.
> But there are millions of machines where this is always true,
> and on all machines this is usually true.
>
> Thus, the failure that happens when the user is not in fact at the
> console will only happen rarely. If it is not a disaster, it may be
> worth ignoring.
>
> And if it is possible to enable the feature through a configuration
> switch, users could enable it on their personal desktop machines.
> They could still log in remotely, but they would not use the floppy
> drive when doing so.
>
> But I have trouble imagining what sort of feature would be based on
> this assumption. How can one take advantage of the assumption that
> the processes that access the floppy are run from the console? Why do
> you think we have to choose between allowing remote logins and
> automatic unmounting of floppies?
I don't consider myself enough of an expert on Linux to judge the merit of
the solutions (and criticisms) posted on L-K. But here's the discussion
which prompted my thoughts.
Anthony Barbachan said:
> Would it be feasable to display a prompt to the user (on the console)
> in this case? (i.e., when the floppy is removed while still in use)
>
> - Save video screen and mode
> - Display critical warning prompt message "Please reinsert disk,
> enter to retry, esc to singal error"
> - After user responds restore the video mode and screen.
Jessie Pollard eventually replied:
> 1. the kernel doesn't know who to contact.
> 2. No one may be logged in
> 3. the console may not be logged in, but other users are logged in over
the net
> 4. If a daemon recieves the message, who gets the output?
> 5. root doesn't have access to the X server
> 6. the person logged in (using X) may not be authorized to answer.
I'm sure GNU/Linux can do anything Windows can, from a development
perspective. But Windows makes a completely different assumption about the
user at the console. Windows requires the operator to be physically using
the hardware; Unix does not. Assuming that the only user on the machine is
at the console simplifies some problems. In this case, we could notify the
user as Anthony described.
In a single-user mode, I think video performance could increase, by having
games write directly to the video buffer (obviously not something which
works over a remote X session). Do we want every game to run faster under
Windows, simply because of the remote-session capabilities of X? And I
wonder if there were other times on L-K when one person suggested a cool
thing to do, and someone else shot it down by saying "it wouldn't work in a
multi-user environment."
However, the more elegant solution to ANY problem would be to solve it for
the multi-user case. In fact, if floppy drives were designed better, they
would work much better in a multi-user environment. I would much rather
solve any issue for GNU/Linux in general, rather than simply with a
single-user desktop version.
chris
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Jun 23 2000 - 21:00:19 EST