[rekonq] quit vs. close

[Andrea Diamantini]

On Friday 19 August 2011 10:10:03 Thomas Zander wrote:
> On Friday 19 August 2011 09.34.52 Andrea Diamantini wrote:
> > So, back to the topic. I have no big pros/cons about the "general"
> > behavior,  that is the quit action closing the current window or the
> > whole app. But, having tested this in these months, I can say I like
> > the optional check changed in the queryClose:
> > 
> > - In rekonq 0.7 we warned the user on window close if he has more than
> > one tab  open. I think that 99% of the users consider this check
> > boring.
> Where does that feeling you have about users thinking its Ok come from?
> 
> In the KDE user research done some years back the result was the opposite,
> the bugreports on konqueror asking for this dialog (also some years back)
> also show the opposite.
> 
> So are you sure that you are thinking of the common KDE users here? To help
> with that question, please notice the 'personas' KDE has created for just
> such a question;
> http://techbase.kde.org/Projects/Usability/Principles/KDE4_Personas
> 
> > - In rekonq master, until 2 days ago, on quit action, if you have just
> > 1 window open it simply closes. If you have more than one, you are asked
> > about  closing all them or simply closing the current one. It just works
> > for me.
> 
> I understand it works for you, makes sense since you are fully aware of the
> fact that all those rekonq windows under water are just the same process
> with one taskId in the computer.
> This implementation detail is not clear to the user of rekonq. She will
> click a link in konversation and get a new window on that virtual desktop. 
> Closing that window using 'quit' does in no way make sense if this also
> closes the rekonq window that was already running on another virtual
> desktop or another X-session or even another activity.
> Especially since other KDE software doesn't behave like that.
> 
> So, in short, it doesn't make sense to me, and from experience I know that
> it doesn't make sense to a lot more people :)

It seems to me you are suffering the same problem you see in my workflow. In 
fact I sincerely trust that the quit = close window habit comes from the 
multitasking design of konqueror.
Let's expose the problem in this terms: if we add the quit action at the end 
of the rekonq menu (in the same way Firefox or Chromium do) the quit action 
now will close one window, while in Firefox and Chrom*, the same will close 
the whole app.
And I'm quite sure this make sense to a lot of more people of the ones used to 
dolphin/konqueror behavior.

> > Please also consider that, as tab history saving will be pushed on
> > master
> > this  morning, the 1% of users who have closed by mistake window instead
> > of tab can resurrect all the others just restarting rekonq
> 
> Thats great! And I'm sure a lot of users will like that feature. I'll try it
> soon as well. :)
> I would disagree with the conclusion that this solves the problem; that
> would imply a mis-interpretation of the problem.
> The problem is predictability of functionality and trusting it to behave
> consistently with the resulting fear of using functionality if this trust is
> broken.
> The fact that a user can recover if the app didn't behave the way he
> expected doesn't change the fear created at the first negative surprise. 
> It just makes it easier to bear.

I agree with this statement. I just don't agree with the fact that in this 
case common people will find the behavior "unexpected".

> I've seen a user get utterly frustrated and he walked out of the room when
> an application unexpectedly exited and he assumed that this meant the last
> 5 hours of work were lost.
> Only an hour later, he was still kind of stressed, did he find out that by
> restarting the application all his work was still there.
> This made him happy, but the real damage was already done. The
> end-evaluation of the application was that he didn't trust it with his time
> anymore. (he pressed save 5 times a minute from then on when he was forced
> to use it)
> 
> Being consistent and predictable is important for user satisfaction. If you
> can't be consistent with everyone else, make sure you go with the 'safe'
> ones so losing work is cut to a minumum.
> 
> I hope this makes the reasons for the behavior a little more clear :)

-- 
Andrea Diamantini, adjam
GPG Fingerprint: 57DE 8E32 7D1A 0E16 AA52 59D8 84F9 3ECD DBF9 730F

rekonq project
WEB: http://rekonq.kde.org
IRC: rekonq at freenode