Re: [PATCH] gnome-vfs: Don't monitor read-only filesystems

[Alexander Larsson <alexl redhat com>]

On Sun, 2004-04-11 at 00:23, George Farris wrote:
> On Fri, 2004-04-09 at 12:30, Alexander Larsson wrote:
> > On Sat, 2004-04-10 at 07:12, George Farris wrote:
> > > On Thu, 2004-04-08 at 10:50, Manuel Amador (Rudd-O) wrote:
> > > > > > No, killing fam was enough.. Fam was maintaining an open fd on the cdrom
> > > > > > drive. Killing it allows unmounting the drive.
> > > > > 
> > > > > Killing fam was enough? Thats like saying "its not an app fault, killing
> > > 
> > > Except that is is seriously unthinkable to ask a user to reboot their
> > > station every time they want to unmount their CD-ROM.  We are all
> > > wanting GNOME "to just work" and then something screwy like fam gets
> > > mixed in.  Bah!  I've been migrating users over to Linux for ages and
> > > it's these little things that completely turn them off.  So, yes,
> > > killing fam and telling it to never start again is the right thing to
> > > do, until it gets fixed.
> > > 
> > > And yes I know how to get the cd to unmount without rebooting but no
> > > newbie and I mean NO newbie will easily figure it out.
> > 
> > Of course its bad that we hang the cdrom unmount and we're working on
> > fixing that, but I'm completely confused by what your mail is about. Do
> > you think its easier to tell users to su to root and kill fam than
> > killing the user process that is actually causing fam to lock the
> > unmount? Thats what my mail was about...
> 
> Sorry Alexander, maybe I wasn't clear, it was late last night when I
> posted:-(  I'm just not clear on why it would be difficult to ignore the
> /dev/cdrom devices in fam.  
> 
> What is difficult is telling users coming from Windows how to find,
> understand and kill the process when they've never had to do something
> like that before.  It's completely alien to them.  Anyway I'm not
> helping get it fixed and should shut up.  I was merely pointing out that
> fam should probably never have been released with this issue.  It has
> caused many phone calls with "how do I eject my *&&*%&$%^$ cd" and all
> sorts of email on various mailing lists.
> 
> We have to fight so many issues migrating users that these types of
> issues should be a no brainer.

And i'm trying to explain, so that people will blame and fix the right
thing, that fam is *not* to blame for this. Do you blame the kernel if
an application has a file open on the CDROM, prohibiting unmount? No,
neither should you blame fam when some app has a monitor open on the
CDROM. Fam is just doing what its told. (And hardcoding fam to not
monitor on CDROMS doesn't solve the issue.)

Fixing this issue does not involve changing fam. It involves fixing the
applications that break unmounting by keeping monitors on directories
that are not actively used.

I do *not* think its a good thing they people can't eject their cdroms.
However, we need to actually fix the problem, not just blame the nearest
app.

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 Alexander Larsson                                            Red Hat, Inc 
                   alexl redhat com    alla lysator liu se 
He's a scarfaced arachnophobic hairdresser who dotes on his loving old ma. 
She's a virginal Buddhist snake charmer who hides her beauty behind a pair of 
thick-framed spectacles. They fight crime!