Xfce Digest, Vol 91, Issue 19

[Edscott Wilson Garcia]

El 23/06/11 02:20, Nick Schermer escribió:
>
>> How long does it take a user to identify a paper in pdf format, when the
>> title is unknown? With Nautilus you have to click and click and click, and
>> then back to the click and click and click.
> And how does rodent identify this file you need?
>

The user is the one who must identify. Rodent makes the process easier
and faster by reducing the number of clicks to zero. For each file to be
examined, a decent size tooltip image will appear. Same applies to text
files. To put the CPU to work, just hover quickly over all files to look
into and then go one by one making the tooltip appear. This feature can
be disabled globally or on an exclusive directory basis.

>> How long does it take a user to run a command in background? With Nautilus
>> you have to click and click and click and then realize you have to install
>> something extra, restart and then click and click and click before you can
>> type (if you can still remember what you wanted to do by then).
> What exactly do you want to run in the background? You mean file
> operations (those run async in another thread in gio) or spawn
> commands (cp/ls/etc)?
>

Wrong question. With a multi-core box, the question is "what *don't* you
want to run in background?" That's specially important when using X,
because X is not thread-safe.

>> How long does it take a user to customize an application to use when opening
>> a certain type of file? With Nautilus you have to click and click and click,
>> and maybe you'll figure it out.
>>
>> Rodent's idea of being fast is allowing the user to do stuff faster, not
>> microseconds to display a label. Nautilus plays in a totally different ball
>> park. Nautilus wants to appeal the greatest number of potential users, most
>> of which do not know what "command line" means.
> Come on, it's not that bad. The difference is that nautilus will store
> the account-wide setting (so all apps using gtk/gio benefit), for
> Rodent you probably have to do it a second time if another application
> wants to open the same mime-type.
>

Sure, it's not that bad. But it's not for me. On your second point, you
are wrong. Rodent uses a account wide setting (and configurable system
wide setting). Other applications are welcome to use them. But that is
bad practice in my book. I hate it when application A changes the
configuration of application B. It reminds me too much of Windows.


>> I could go on and on and on. You might say that not all users want skim
>> through dozens of scientific pdf articles, or run simulation programs in the
>> background or use an application other than the officially gnome santioned
>> ones. And you are right. Rodent is not for everybody.
>>
>> More numbers? On initial installation Rodent will not appear as fast as it
>> actually is. On virgin installs, disk and memory caches are empty. Once you
>> are using Rodent in a realistic way (not stress testing under ideal
>> conditions), these caches bring the numbers down significantly.
> That's the same for every application.
>
>> CPU usage? Rodent is built around a multithreaded design. CPU speed has
>> reached a plateau and the only way go faster is to do more work in parallel.
>> If you do not have a multicore processor, Rodent's design will make is
>> slower than any other program doing the exact same thing. But multicore
>> processors are here to stay.
> Threaded as in spawning processes or in multiple threads in the same instance?

Both, and the combination of both.

>> At my office I have "common Joe" box, and it has 4 cores in the chip. Last
>> time I looked (at least a year ago) AMD was attaching 12 cores to the chip.
>> Does 5% of one, hurt? Not for me. After all, that's one of the reasons why I
>> have a computer, to put it to work.
> It just a sign of incorrect coding. It's not about the cpu working.
> but about power consumption: battery life and a greener world.
> Companies like intel try hard to reduce power, if your app does not
> sleep, it nearly impossible to do that.
>

Nope. The application does go down to very low CPU levels. But this will
not happen while the user is moving the mouse around. Mouse movement
gives it the clue to what to attempt in predictive computation.

Weird how you mention a greener world after you considered the bicycle
stuff as crap. A years worth of bicycle commuting will reduce more
greenhouse gas emission than leaving your computer off the same period
of time. And it's downright fun.


>> The important number here is RES. Considering I have 8GB on this
>> "run-of-the-mill" box, what's 35 MB on the  rodent instance? Answer: tiny.
>> Even firefox does not have me worried. At this time I cannot compare with
>> Nautilus because, well, I don't even have it installed. Why not?
>> Dependencies.
> 8GB is not normal. Netbooks have 1gb and _a lot_ of people have less then that.

My netbook has 1gb and a single core. It is the bottom line for Rodent.
Rodent's performance on this layout should be acceptable (to me, at
least). Performance gains on better hardware should be the rule, because
hardware is getting better all the time. These are Rodent guidelines.

>> What about dependencies? If I do an "emerge -p Nautilus" on this Gentoo box,
>> I would need to install no less than 38 packages before it even considers
>> installing Nautilus. Everything from libsoup to gnome-desktop.  Gvfs brings
>> in 34 packages by itself (practically the gnome desktop). That's too much
>> for me, especially when I compare it to the emerge for Rodent.
> I agree on that one, but maybe it's better to choose a more
> competitive file manager to compare with.

I guess you're missing the picture here. Rodent is not out to compete
with any other filemanager. Rodent is there to fill a void. Rodent is
there for all the users who at least once stared at the filemanager on
the screen and thought "Geez! I'm not *that* stupid!". I'm specially
irritated by filemanagers that create folders and tell you what should
go in them (videos, pictures, documents,...).

>> The bottom line is that Rodent is small, fast and powerful. If your
>> definition of small, fast or powerful is different, that's fine too.
> I understand what you mean and I'm glad there are file manager around
> like Rodent, good to poke tradition once in a while. But if you say
> it's better you shouldn't only use the negative points of the *other*,
> but also show your workflow, so people can decide for their own if it
> is better for them.

Better is a relative concept. If you ask which soccer team is better,
you will never end. I've got my wife set up with Ubuntu and Nautilus.
She's entirely happy.

Rodent 4.6.8 will have the ability to view the process tree as a file
structure (as everything unix) which allows for greater control of
background processes so you can click a signal off to any process.
Clicking off a SIGSEGV to a deadlocked thread is very useful for someone
like me. But is totally Chinese to my wife or the average Nautilus user.
Yes, I know I can pipe ps to grep and try to figure it out, but I'm in a
hurry.


> The question how Rodent is better remains a bit vague for me, but
> that's not entirely fair since I know a lot better how Thunar/Nautilus
> work.
>

Yes, that's a problem we all have. When we are milimeters close, a
perspective is hard to get.

Thunar is better than Nautilus because it does not deny me a terminal on
initial startup. But I'm sure Nautilus fans will argue the contrary. Shrug.

-- 

cheers!

Edscott