[Thunar-dev] Spatial or not-spatial?

[Brian J. Tarricone]

Ok, I just read through the 20-odd new messages in this thread, so I'm 
likely replying somewhat out of order.  Here goes...

Benedikt Meurer wrote:

> Jasper Huijsmans wrote:
>
>> Important for me is that there is only one window. I know where my 
>> files are, so I just go there. If I want to move/copy it somewhere 
>> else I open a new window and drag things over. This is most likely 
>> not the way most users work.
>
Exactly.  I don't want lots of windows open (if I wanted that, I'd just 
open a bunch of terminals, like I do now).  I want one window that 
allows me to browse (navigate, if you will) my filesystem.

> I guess most users would just right-click the file, select 'Cut file', 
> use the treeview (or probably the up/back buttons) to navigate to the 
> destination folder and choose 'Paste files' if they are used to/forced 
> to use a browser like file manager.

Or, assuming a sane DnD model (where the treeview items expand when you 
hover over them, and the view scrolls when you hover over the top and 
bottom of the list), you drag and drop files where you want to copy/move 
them.  Frankly, Windows Explorer has it pretty much dead-on here.

>>> Well, as said, personally I'd prefer the spatial view, but as Auke 
>>> said, its not about our personal opinion...
>>
>> Hehe, you start to sound just as undecided as the rest of us. Like I 
>> said, I see advantages to both, but i really don't know which one to 
>> choose.
>
> I'm not undecided; I've made my point clear.

Agreed.  I'm just about as far from undecided as you can get, just in 
the opposite direction as Benny.  I see the value in a spatial file 
manager (esp for new users), but 1) that's not how I work, and I don't 
want to work on or use a file manager that doesn't fit my workflow, and 
2) I think there are some serious usability problems (such as the large 
number of windows you tend to end up with when you work spatially) that 
haven't been adequately addressed.

> In my opinion, Xfce should focus on ease-of-use and simplicity in the 
> future (I'm sure everybody remembers that *useless* debate on 
> xfce4-dev); we don't need another geek toy for the Unix/Linux desktop, 
> we have already a lot of complex file managers out there. I don't care 
> if a bunch of geeks don't use the final Thunar version (in fact I 
> doubt that I'll ever use a file manager on a regular basis myself), if 
> we can make the average user happy with a simple file manager in Xfce 
> 4.4.0.

That's where I think we have a bit of a difference.  Benny, since you're 
getting paid to work on Xfce, you'll write a file manager that you won't 
necessarily use.  I won't.  The things I hack on are the things that I 
use.  When I stop using them, I stop hacking on them.  Occasionally I'll 
do a one-off thing for a friend because I think it's an interesting 
problem/challenge, but I'll rarely spend my free, unpaid time working on 
something I don't care to use myself.

> "Beginners are, by necessity, task-oriented. It is very hard for us 
> techies to understand beginners, not because we know the stuff well, 
> but because we're professional learners who have evolved a highly 
> efficient personal learning methodology without even thinking about 
> it. Our tendency is to present facts and examples. We structure 
> information the way we would like it, not the way the user needs it."
>
> Its really necessary to keep that in mind.

I fully agree with that, but, as I said above, I want to create 
something that *I'll* want to use, first and foremost.  Well, maybe not 
first and foremost, but it's up there near the top of the list.

> Oliver (my boss) made an interesting point recently about the file 
> manager design:
>
> Imagine the WWW, its hierarchically by design:
>
>  org
>   |- xfce
>   |   |- thunar
>   |   |   | - /wiki
>   |   |   |    | - /dokuwiki.php
>   |....
>   |- gnu
>   | ....
>  ....
>
> and so on. Many people tend to think of the WWW organization as a 
> complex graph, but if you think twice, you'll discover, that it's 
> really just a simple tree (with *lots* of nodes).
>
> Now, if you are looking for information, e.g. you are looking for 
> information about 'graphical file managers for Unix', then you could 
> of course traverse the `WWW tree' to search for matching documents. 
> But nobody would do that - really, ask yourself. :-) Instead you'll 
> most probably query google (or msn, or whatever) and check the results.
>
> The question is: Why? The WWW is organized as a tree, why not traverse 
> it?
>
> The answer is simple: It takes too much time, because there are too 
> many nodes. Thats why you'd use a query-interface, rather than the 
> tree-interface.
>
> So, lets have a look at or file system. Of course, nobody has the WWW 
> on his file system at home - "has anybody seen my internet backup?" - 
> but the amount of data on home desktops is ever growing, with no end 
> in sight. And the answer most people gave me here recently is that you 
> need the tree-interface to master this amount of data.
>
> Hm, ok, lets recall, for the WWW you use the query-interface because 
> thats the way to master a huge amount of data, but for your file 
> system you use the tree-interface because that's the only way to 
> master a huge amount of data - btw, to make sure nobody gets stuck on 
> the terminology: "data" = information, no matter if that information 
> is presented by local files, by static html page, by table rows in 
> databases, etc.
>
> That doesn't make sense on the long-run (in *my very own* opinion). 
> Instead it sounds like some old bad habit that doesn't want to die yet.

And I'll reiterate that I think the WWW is a terrible analogy to use 
here.  Sure, domain names (followed by web pages in that domain) are 
organised somewhat hierarchically.  But there are a few places 
(important ones, IMO) where that analogy breaks down:

1) The domain name (hostname?) / file path disconnect.  org.xfce.thunar 
makes thunar a sub-item of xfce, but wiki really has nothing to do with 
that hierarchy.  It's an implementation detail of the wiki, which is 
(IMHO foolishly) exposed to the user.
2) In the beginning, there were no search engines, and people got to new 
websites based on a) word-of-mouth, b) hyperlinks between them.  There's 
no analogue in the filesystem world to hyperlinks (symlinks are a bit 
too primitive, and can't be embedded in data reliably).  Search only 
became popular when the sheer volume of data on the WWW was too much for 
a user to reasonably navigate via the original method.  I think you'd be 
hard-pressed to say that a user needs a search feature for their file 
system just as much as we need a search engine for the WWW (not saying a 
filesystem search tool isn't useful).
3) Domain name hierarchies are inconsistent between domains.  
org.xfce.thunar means that Thunar is a sub-project of Xfce.  
org.spuriousinterrupt.kelnos means basically nothing - it's only 
apparent that kelnos is the name of my home machine to someone who knows 
me (and even then...).
4) Filesystem hierarchies are inconsistent between webservers, and may 
even be inconsistent on the webserver itself.  As I said before, 'wiki' 
on the thunar server really has nothing to do with anything the user 
needs to see.  Also, I think many people are less-likely to organise 
files on the webserver because 1) they believe that the files will be 
navigated via hyperlinks, and so the locations on disk aren't really so 
important, or 2) they believe that the files will be found via a search 
engine.

Anyway, I don't want to beat this to death, but I really hate the use of 
analogies when they don't really make sense.  I love analogies, but only 
when they work.  IMO, this doesn't work.

> And remember, its for the user (!= 24/7 geek), not for you (well, 
> maybe its for you as well, but that doesn't matter).

As I said before, from my perspective, it's not for the user.  It's for 
me.  I don't get paid to write software, and I'm mostly only going to 
work on things that I'd want to use myself.  That's not to say I'm not 
going to add features to make it more accessible and usable by 
less-advanced users.

> Ok, even that appendix is too long. Hm, too late already...

Yeah, I think we all just have too much to think about/say on this 
topic.  But I think it's better to have too many ideas than too few.

    -brian