Re: [Usability] Re: Tabs

["Liam R. E. Quin" <liam holoweb net>]

On Mon, 2005-06-27 at 23:42 +0200, Reinout van Schouwen wrote:
> Op Mon, 27 Jun 2005 19:13:57 +0200, schreef Thilo Pfennig:
> 
> > I think important is something like if somebody closes a window and there
> > are many tabs open, that the application asks if the user really wants to
> > close all tabs.

An alternative is to provide Undo.  Consider a text editor that asks if
you really want to delete a line of text, vs. one with undo :-)
On the other hand, while I was typing this message, a dialogue box came
up asking if I really wanted to discard the message.  I've no idea
why, although I might have pressed control-w to delete a word... *tries*
yes, "delete word" tries to close the window, stupid binding.  If the
window simply vanished, how would I know if the mail had been sent, or
if I'd saved the message for later, or if I'd deleted it?  So the
dialogue is needed.  Better might be to animate the window going into
the /dev/null abyss and a sound of a scream, plus a text log saying
what had happened, e.g. in a scollable status section in the mailer's
main window.  Then if I wasn't looking at the screen, I'd be able
to figure out what had happened and press Edit->Undo or whatever.

If I'm not looking at the screen, I'm likely to press space or even
enter not noticing a dialogue box happened.

> Not necessarily. The important thing is that data loss is prevented.
> When the user presses the close button / menu item on a window, why
> second guess that maybe she actually meant to close one tab? Only when
> there's unsaved data, a dialog may appear-- but it would have appeared
> too when not having tabs. Plus, in a web browser at least, there's
> always the history that allows you to go back to accidentally closed
> webpages.

So, back to tabs...  if I close the wrong browser window by mistake,
up comes "are you sure you want to close this window?" and I say yes
because I have not realised I clicked on the wrong "X".  Then I go
"oooh shiiit" because those windows had been up for days,  I was
saving them in tabs until I booked my hotel for a trip.  So maybe they
are in my history like you say and maybe not, depends if I cleared
history, depends how long my history lasts, no?  If they are in my
history, how am I going to find them?

So for your statement "they are in your history" to be anything
other than useless :-) the history needs to have a "window closed time"
field I can sort by.  I note that for a Web browser, Undo Close Window
still doesn't really work because there's no single main window.

I restarted my Web browser (galeon) this morning and it started
loading a dozen or so tabs.. it wanted a username/password, and hung
until I typed it.  No idea which tab, which URI, not even 100% certain
it was Galeon asking, but it seemed likely.  So there are other issues
with tabs, and with associating dialogues/questions/errors with them.

I tried to imagine a Web browser that didn't have these problems, but
it came out fairly radically different, and maybe not better.  E.g.
a "main" window where you can type a URI/IRI and it'll open in a
new window, or you can drag it onto an existing window.  A URL bar
on a window then becomes read-only and means "this is what I am
showing".  The main window can then have a place to undo a close
window, to get at history, to configure preferences... I suppose more
like Gimp, although I'm not sure I want to hold Gimp up as a UI to
emulate, as it's addressing some very complex editing tasks.

Let's not dismiss problems with tabs as "oh, it's all solved by history"
though, when in fact it isn't :-)

It's difficult to provide pervasive Undo on a desktop, because we are
stuck with a file system that doesn't have Undo and, ultimately, we
interface with a real world that doesn't have undo.  Putting your
shoes back on isn't the same as to undo the act of removing them,
because you don't forget that wonderful feeling of relief and
comfort :-)

So the abstraction is a little leaky at the edges, and some of the
questions that are coming up seem to me to be results of adding tabs
as a convenient hack to programs -- I'm not primarily thinking of Web
browsers here -- e.g. the interaction between "spatial" and "tabbed"
is clearly weird -- should a tabbed nautilus window jump around the
screen when I change tabs, to reflect correct positions? :-) -- and
the tab also confuses the distinction between document and window.

Tabs have some (but not all) of the problems that MS Windows "MDI"
had, but with a lot more flexibility & more benefits.  The question
of where to put the [x] is an indication of this...  

Best,

Liam

-- 
Liam Quin, W3C XML Activity Lead, http://www.w3.org/People/Quin
Pictures from old books: http://www.holoweb.net/~liam/pictures/oldbooks/
IRC (chat) programs: www.ircreviews.org/clients/