On Thu, 05 Mar 2009 15:36:13 -0800, Darren New wrote:

>> I think that's a little different than the scenario that was discussed,
>> though - "I'm going to ask you for permission, OK?", followed by "I'm
>> asking you for permission" is a little confusing.
> 
> I suppose. It's the same sequence of events as the Unix mechanism,
> except the Unix mechanism doesn't give you the same kind of prompt. It
> makes you start over, instead.

Well, not really.  The Unix mechanism is saying "you tried to do 
something that you don't have permission to do".  Then the user tries 
with elevated privs.

The Unix mechanism isn't saying "I'm going to ask for permission to do 
this".

For example, if I run yast2 -i on my openSUSE box, it's going to prompt 
me using gnomesu or kdesu for the root password (and the dialog explicity 
says "root privileges are needed to perform this action").  It doesn't 
try with lower privs (unless I type too quickly, I need to file a bug on 
that) and then say "hey, I need to ask you for permission to do this, 
back in a second" and then a dialog prompts me.

>> It would be better for the second prompt to just explain about it
>> rather than have an additional prompt.
> 
> Maybe. On the other hand, this way they can make the code small and
> clean, without having (perhaps) a problem with getting to the help file
> and so on. I.e., if you made it so the prompt could tell you everything
> you needed to know, it might wind up needing things that only the
> unprivileged session can get to anyway.
> 
> I'm not saying it's the best way to do it. I'm just saying it's not as
> bad as people make it out to be, because such a warning is actually more
> consistent than randomly popping up a box asking for the admin password.

Well, heaven forbid we train users on how to do things and what things 
mean. ;-)

Seriously, sometimes it's a wonder that they ever figure out how to use a 
word processor or spreadsheet.  The apps don't need to treat them like 
idiots, there's no reason for the OS to do so either.

My mom - who has been kinda the "poster child" for this type of thing for 
me - was easily trained on how to know when it was appropriate to accept 
an outbound connection when her firewall app (the name of which escapes 
me at the moment) asks if the app should be allowed to do this or not.  
She's in her 70's and not the most computer literate person on the planet 
- prior to getting a PC, the most she did with a computer was play Pac 
Man on it.  Seriously.

>> The better design would be for the prompt to tell you at the time it
>> comes up, not to warn you "I'm going to prompt you in a second for
>> permission for this".
> 
> Why would that be better?  I think I've already explained why the
> two-prompt is better. I think your mechanism is only better if you're a
> nerd and you already know when you're doing something that's going to
> trigger the prompt.

Well, I disagree.

> Explain to someone who doesn't understand computers how to tell when
> it's OK to answer yes to the prompt.

As noted above, have done so.  It's not rocket science.

Jim

Post a reply to this message