LWN: Comments on "Did Linus Jump Too Soon? (Linux Journal)"

Did Linus Jump Too Soon? (Linux Journal)

mjthayer — Fri, 06 Feb 2009 08:31:58 +0000

I have actually looked at GoboLinux before. They certainly take a good shot at structuring a Linux filesystem like OS X or NextStep. It is still not quite good enough for me though :) I think what I would really like would be something that would fit in well to existing Linux systems, rather than trying to start afresh (that always feels nicer, but is usually doomed to remain a niche). Fitting in with existing systems would of course require a lot of thought, much talk, many compromises and a delicate sense for diplomacy. And of course GoboLinux just look at this from a filesystem point of view. Other things, such as process management, system configuration and service management could also do with some restructuring. Unfortunately, we don't have something as nice as OS X's filesystem to copy here :)

Did Linus Jump Too Soon? (Linux Journal)

DarthCthulhu — Thu, 05 Feb 2009 19:51:58 +0000

"To tell the truth, what I would love would be a desktop system that is easy to use because it is well structured, not because it hides things from the user."

You might want to look at GoboLinux (http://www.gobolinux.org/).

It's a distro with a completely reorganized file system tree that is MUCH better than the old *NIX way. The best part is, through symlink magic, you can still look at the filesystem in a legacy way if you (or any processes) want to. Everything works really nicely and there's no need for complicated repositories; adding or removing software is done via "recipes" which are simple scripts. If necessary, you can even just manually add or delete things... the recipes are just nice helpful things which make setting and unsetting the various symlink magic easier.

decision for community

Arker — Thu, 05 Feb 2009 05:30:07 +0000

I didnt call it 'crap' and I didnt sling any mud. Just straight up observations from an interested party (a KDE user who's been using Free Software and boosting for it since '94.)

While your explanation for what happened makes sense, it doesnt make the KDE developer community look any better. What you're saying is that you have to mislabel your development releases to get developers to work on them. I'm very sad to hear that. I can't help but think that your first goal should be to back up a bit and work on the flawed assumptions behind the response. Developers shouldnt have any problem working on alpha/beta releases - another name for them is 'Developers releases' after all.

Linus - New Interview More Good Copy

jospoortvliet — Wed, 04 Feb 2009 13:56:13 +0000

Well, fedora has reasons to do what they do - they want to advance the state of FOSS, and by pushing they latest & greatest they hope to do that. You can agree or disagree with that strategy but as long as users make a clear decision to join that or not, I think it's OK.

I agree I wouldn't use it either, btw, but that's not the point.

Linus - New Interview More Good Copy

Arker — Wed, 04 Feb 2009 00:57:58 +0000

One more good reason to avoid Fedora like the plague.

I'll probably never understand why anyone would use it, but meh, his choice.

decision for community

sebas — Tue, 03 Feb 2009 14:48:07 +0000

The "borked" versioning you describe has everything to do with the overall
quality of KDE at the point of 4.0. Here's my personal, frank analysis of
it:

- Most of KDE 4 was ready for a sensible .0 release at that time
- Plasma was lacking features and stability at that point
- The decision to ship KDE 4.0 was a decision for the community and had
relatively little to do with users

So what happened then? Plasma, being the most visible component in the
desktop was of course immediately found to be lacking. Interpolation made
that a "KDE 4 is crap" (note how both the state of most applications, the
development platform and the fact that 4.0 is the kick-off for the new
KDE4) is ignoredin this. This is from a user's point of view.

From a community point of view (which almost nobody seems to think about
unfortunately), in 2007, one of the biggest problems was to get the
community onto KDE4, to start porting apps, to stop tinkering with
technology and to make KDE4 releasable, this is commonly called "release
mode", and it was one of our biggest challenges from changing the pillars
into an integrated whole. We started releasing Alphas, Betas
and RCs then, and it indeed worked. Our people understood that KDE4.0 was
to be released soon and got on it, started porting apps, worked on fixing
bugs and polishing. Our people got into this release mode and have been
frantically fixing KDE4 for months. Some modules, for example the KDE games
module, but also KDE Edu release-quality well before 4.0, others were
slower. Plasma is one of those slower components. It had been started quite
late in the process, mostly because we needed some stuff fixed in Qt, and
because it would layer on top of almost anything else so technically, it
*had* te be done last.
At some point though, we would have to release KDE 4.0 (you cannot warn
forever, you cannot keep people in frantic release mode forever), and we
decided to do it although Plasma wasn't ready for prime time -- it was
basically usable though (i.e. I could run an editor, a konsole and some
other programs, even if it wasn't the most fancy thing
around). It also happened to run stable, so I assume it was at least good
enough for some people. It would have been unfair to non-Plasma developers
(which there are quite a lot within KDE) to hold everything back until
Plasma is good enough for $group_of_users, and it would've held back other
processes, such as the move to KDE4 among 3rd party KDE apps, fixing of
issues in Xorg, various video drivers, and other components throughout the
stack.

So did that work? I would say yes. The KDE4 frameworks are quickly
maturing, we're seeing KDE3 apps being ported all over (with the more
complex ones such as Amarok, KDEnlive having their first KDE4 release and
others such as Digikam, and K3b nearing readiness. The number of commits
as well as the number of hackers on our codebase is steadily increasing,
and many other things in our community making sure we'll scale well in the
coming years. So 2008 has been a very good year for KDE, despite all the
complaining.

How did we deal with the 4.0-backlash? This is in my opinion an important
point, because it says a lot about how we deal with feedback. What indeed
happened is that many people (especially inside the Plasma team) put aside
their grand plans in order to work on things that users were missing in KDE
4.0 compared to KDE 3.5. This is mostly completed with KDE 4.2

So the trick users into testing is a quite superficial analysis, and simply
not true. It has been a decision that largely underestimated how users
would react. We did assume our messaging was fine, reflecting our pride of
having gotten something as big as 4.0 out, and even thought users were more
sensible and would "get" it. That's obviously at least for a very vocal
group not true.

While you can have any opinion you want on the 4.0 release, repeating that
4.0 was crap deceiving is only part of the truth, and not the forward-
innovation and "balls" are not appreciated, and that you'll burn yourself
with those "great plans". I'm personally glad that most KDE developers keep
working despite this mudslinging, and I'm confident that the way KDE4 is
going shows its promise to more and more users.

Did Linus Jump Too Soon? (Linux Journal)

mjthayer — Tue, 03 Feb 2009 14:13:32 +0000

As I said, I wasn't making an argument that Apple's AppFolder concept is appropriate for Linux (although I'm not saying it couldn't be either, in a sufficiently rethought-out form :) ). It was just an example of how a user interface can present things simply without dumbing down the user - in effect the complexity is layered, and you can go down a layer if you need to take a deeper look. Whether the same principle of layering the interface and the complexity could be made to work in other situations, like controlling system services or components of the sound system or kded (whatever that is - it has died on me too, but I also never quite understood what that meant...) or whatever, I don't know.

Did Linus Jump Too Soon? (Linux Journal)

jospoortvliet — Tue, 03 Feb 2009 12:16:52 +0000

Of course my comment was as much targeting the parent of yours than your own comment. I disagree with your statement about cautious versioning (I don't see a big diff between FOSS and proprietary versioning, I even think proprietary software is generally a bit more cautious), and you can't disagree that the parent was pretty - unfriendly.

Then again, maybe I shouldn't respond like that. It's just that I've been writing extensive responses to flamy posts like the parent of your comment for what, a year now? Sometimes one gets tired of answering the same uninformed questions again and again.

Did Linus Jump Too Soon? (Linux Journal)

dlang — Tue, 03 Feb 2009 11:51:37 +0000

a large part of this is the multi-user heritage that Unix has (and Linux inherits)

part of this heritage is that a given app should only need to be installed on a system once, and there should be a set of system-wide defaults.

while there is defiantly a place for single-user focused setups nowdays, the vast majority of *nix systems out there are still structured for multiple users. so any distro trying to do something different will be faced with a large body of existing code that won't do things the 'new way', and any software trying to package itself in the 'new way' will have to decide how to work on all the existing distros that work the 'old way'. trying to work on a system that had some packages one way and some the other could be far worse than either by itself.

And while many geeks have at lest one system per user nowdays, the 'average home system' benefits enough from multiple users that new systems (including windows) are putting more emphasis on supporting multiple users than they have in the recent past

I don't use OS-X, so I don't know how it deals with these issues (other than the fact that it doesn't bother with compatibility with anything else)

Did Linus Jump Too Soon? (Linux Journal)

dlang — Tue, 03 Feb 2009 11:37:50 +0000

to be fair you did make your comment in a thread primarily about the KDE 4.0 release. it would be hard to take your comment as anything other than further criticism about that release in this context.

for those who are continueing to bash the KDE 4.0 release a year later, please provide the time machine so thta the KDE developers can go back and fix it, since it seems apparent that nothing else will satisfy many of you.

note tht I am not a KDE developer, I'm a user who has been using KDE4 for a fair chunk of the past year and, while I have seen things that were nice when I used 3.5, I still choose 4.x over gnome when forced to use one or the other.

Did Linus Jump Too Soon? (Linux Journal)

mjthayer — Tue, 03 Feb 2009 11:26:45 +0000

To tell the truth, what I would love would be a desktop system that is easy to use because it is well structured, not because it hides things from the user. A case in point - compare applications on OS X to applications on a Linux distro. The OS X ones are organised as a single directory containing all files needed for the application. The user sees that directory in their file browser, clicks on it and the application starts. They can remove it by deleting the directory. If they need to "look inside", they can open the directory instead of just starting the application. On a Linux system, the applications are scattered through the /usr tree, which most desktops hide as far away from the user as possible. The user is shown an icon in the start menu which represents that collection of scattered files, and can remove the application by searching for it in the package manager. Quite a difference.

Disclaimer: that was meant as an illustration of a point, not to say that Linuxes should be using AppFolders (they wouldn't easily fit into the Linux world for a variety of reasons) or that Apple does everything right.

Did Linus Jump Too Soon? (Linux Journal)

cantsin — Tue, 03 Feb 2009 11:21:08 +0000

I would prefer if you reserved your kind of response to forums like
Slashdot. In the posting you replied to, I did not mention KDE at all, but
just made a general observation about a tradition and more recent practices
of versioning schemes in free software. - Since you are apparently a KDE
developer, you are not helping your project, but on the contrary reinforce
a perception that it has communication issues.

Did Linus Jump Too Soon? (Linux Journal)

jospoortvliet — Tue, 03 Feb 2009 09:12:38 +0000

Oeh, yeah, you're so right. All foss software uses numbering schemes like Wine and Google. And KDE 4.0 was only released to trick users into testing it (so we could ignore the bugreports and make fun of them).

Linus - New Interview More Good Copy

jospoortvliet — Tue, 03 Feb 2009 09:10:46 +0000

He did say he uses fedora, right? That's the issue, fedora forced KDE 4.0 on its users by not providing 3.5.x packages.

Did Linus Jump Too Soon? (Linux Journal)

cantsin — Tue, 03 Feb 2009 08:26:59 +0000

It should be added that Free Software traditionally has a quality reputation for cautious version numbering schemes. Fine programs like mutt carried 0. version numbers for years. Even today, mplayer, elinks, aspell and inkscape, for example, haven't reached 1.0 yet. In the world of (proprietary) web applications, Google is setting a high standard by calling perfectly usable service "beta". It would be sad if Free Software traded these quality standards for those that are rather typical for the "Shareware" cultures of Windows and Mac OS.

Did Linus Jump Too Soon? (Linux Journal)

b7j0c — Tue, 03 Feb 2009 05:07:42 +0000

journalists love crap like this because its easier than writing a real article. just take something noteworthy from the echo-chamber, throw some cheap contrarianism on it (i.e., linus is wrong), and voila, instant pageview fodder.

Linus - New Interview More Good Copy

Arker — Tue, 03 Feb 2009 01:42:09 +0000

The GNOME mentality is bloody annoying, I agree.

But KDE 3.5.10 is still very usable. Why not just stick with that until KDE 4 is ready?

Did Linus Jump Too Soon? (Linux Journal)

Arker — Tue, 03 Feb 2009 01:36:04 +0000

The problem is that your versioning system is borked. 4.0 was, at best, a beta release (and by some opinions alpha.) A .0 release, unless you're MicroSoft, is understood to NOT be a beta release. What was released as 4.2 should have been labeled 4.0rc1. By adopting deceptive, MS-like versioning numbers you set yourself up for massive backlash, which materialised on schedule.

At the same time, distributors should have caught this and for the most part did not (Slackware being one exception, I think I heard SuSE was another.) In Linus' case, he inexplicably insists on running Fedora. Fedora promptly replaced stable KDE with alpha-beta releases, whether in keeping with their philosophy as a 'test-bed' distro or out of confusion engendered by the deceptive version numbers mentioned above I dont know. Whatever the motivation, the outcome was the same - he found stable usable KDE replaced with mislabeled alpha-beta level new KDE and had to switch to something that worked. That something just happened to be Gnome.

This can be addressed at two levels. You (throughout this post I'm using 'you' to refer to you as a KDE developer, and more broadly to you as KDE Developers collectively) can and should reverse this policy of using a marketing versioning scheme and go back to something that accurately describes the release. You can also try to lobby distributions to quit screwing up. I suspect you'll have more luck with the former than the latter.

Did Linus Jump Too Soon? (Linux Journal)

dkite — Mon, 02 Feb 2009 21:03:05 +0000

You must live somewhere warm.

Go outside is the difficult part around here. I want to know if I want to go outside.

Derek (who looks that his home thermostat with outdoor sensor to make such decisions)

WWLD? (was: Did Linus Jump Too Soon?)

tnoo — Mon, 02 Feb 2009 11:59:39 +0000

Try xmonad.

Linus - New Interview More Good Copy

roblucid — Mon, 02 Feb 2009 10:19:42 +0000

http://distrowatch.com/weekly.php?issue=20090202

LT: Since I'm on Fedora, I got hit by the (bad) transition to KDE4, and as a result I've been using GNOME for the last year or so. It's still somewhat painful, more so when I'm on my laptop, mainly for the same old reason: you cannot fix the mouse buttons in GNOME. (The reason this hits me more on the laptop than anywhere else is that most laptops only have two buttons, making the middle-button press much harder. And middle button is what you need for the 'send to back' window action.)

I wrote the patch (including even the graphical configuration management), I sent it in, and it got rejected as "too complicated for users". Frickin' idiots (and I'm not talking about those alleged users).

But right now, KDE is worse. I'd like to explore alternatives, but if you've followed my answers this far and are perceptive, you'll probably already have figured out that the programs involved aren't on my list of things I care about that much.

I'm well known for disliking GNOME, but it's not the "using it" part that I dislike as much as the apparent mentality of the GNOME people who think that all users are idiots and then limit what I can do with it for that reason.

Did Linus Jump Too Soon? (Linux Journal)

janpla — Mon, 02 Feb 2009 09:58:36 +0000

Does it lack that kind of things? I personally am glad to hear it.

I recently changed away from GNOME for that very reason - what I do mostly on my computer is WORK, and I found that in GNOME, before I could work I'd have to get rid of a load of silly nonsense.

What I like about KDE is that the developers don't seem to be bent on taking away features that are useful to people, even if it is only a minority. The GNOME people are too nannyish; they simplify things beyond what makes sense and want to tell users how to do their work.

Did Linus Jump Too Soon? (Linux Journal)

janpla — Mon, 02 Feb 2009 09:48:23 +0000

> Presumably those people live in places where the weather is predictable

Oh, come on, if you want to know about the weather, all you have to do is go outside or find a window. Primitive, I know, but it works.

WWLD? (was: Did Linus Jump Too Soon?)

debacle — Mon, 02 Feb 2009 09:33:04 +0000

What would Linus do? I don't know and I don't care. I'm thinking about switching from XFCE4 to LXDE on my netbook. What would you do?

Did Linus Jump Too Soon? (Linux Journal)

callegar — Mon, 02 Feb 2009 08:05:13 +0000

On an updated Intrepid with latest kde 4.2 I get

kbluetooth4
kbluetooth4(7927) Solid::Control::ManagerBasePrivate::loadBackend: Backend loaded: "BlueZ"
kbluetooth4(7927) Solid::Control::BluetoothManager::buildDeviceList: UBI List ("/org/bluez/hci0")
kbluetooth4(7927) Solid::Control::BluetoothManagerPrivate::findRegisteredBluetoothInterface: findRegisteredBluetoothInterface "/org/bluez/hci0"
kbluetooth4(7927) Solid::Control::BluetoothManagerPrivate::findRegisteredBluetoothInterface: Creating New Interface "/org/bluez/hci0"
kbluetooth4(7927) Solid::Control::BluetoothManagerPrivate::findRegisteredBluetoothInterface: Calling Backend to Creating New Interface "/org/bluez/hci0"
kbluetooth4(7927) Solid::Control::BluetoothManagerPrivate::findRegisteredBluetoothInterface: BackendIface created
kbluetooth4(7927) KBlueTray::onlineMode: online Mode
kbluetooth4: symbol lookup error: kbluetooth4: undefined symbol: _ZN5Solid7Control16BluetoothManager8securityERK7QString
kbluetooth4(7926): Communication problem with "kbluetooth4" , it probably crashed.
Error message was: "org.freedesktop.DBus.Error.NoReply" : " "Message did not receive a reply (timeout by message bus)" "

Not that nice.

Also, KDEPrint shows some printer options twice and the "kprinter" application is gone, so there can be no printer integration for legacy applications (e.g. no way to provide a dialog to change resolution or duplex mode on older X11 apps).

My point is that while the desktop look is improving and its speed is reasonable, some hidden core functionality is somehow lacking attention and there is less information to understand what is going on behind the scenes, so it has become almost impossible to "play" with the platform and to understand what is going on when unexpected behavious arise (Kded dying with a "kded has aborted" is not very informative and there is no docs to find out what kded does at startup, what files it looks at, etc.).

In latest times, a lot of activity has gone on wrt the desktop: new abstraction layers (solid, DBus), new ways of managing sound (pulseaudio, phonon), new DE paradigms (plasma). In all this there is an expectation that everything shuld just work, so that the user is not expected to know /how/ it does so. But one of the things that made old UNIX fun was the tons of man pages (every single command or config file had one) that let you understand what was going on.

All in all you get a feeling that "you are loosing control", you just need to wait for new versions to come in and you cannot anymore adapt your toy to play nice with the legacy stuff that you still have around.

Did Linus Jump Too Soon? (Linux Journal)

ekj — Mon, 02 Feb 2009 07:32:45 +0000

http://www.yr.no/place/Norway/Rogaland/Stavanger/Stavange...

Amazingly clueful service from the meterological institute of Norway. The data is available as XML, under stable, maintained urls. And under a license which is pretty reasonable. (You have to display "Forecast from yr.no" on publicly visible sites using the data, and you are not allowed to change the content, i.e. add 2 degrees to the temperature at your beach-resort to make the weather appear nicer)

Sadly, they only deliver XML-data for the locations they themselves collect data for. For licensing reasons, they don't for locations abroad, so though they cover the world, they're unusable as a data-source for locations outside Norway. Pity.

They do have several million locations, certainly a whole lot more than "one per city", and the granularity is hourly forecast for the next 24 hours, "morning", "mid-day", "evening", "night" for the next 72 hours. (I get 549 hits for "London" and 86 hits for "Oslo", for example)

Did Linus Jump Too Soon? (Linux Journal)

rsidd — Mon, 02 Feb 2009 06:46:07 +0000

KDE 4.2 does have a weather applet. I couldn't find radar pictures, but I think those are useless eye candy, except for professional weathermen.

briefs or boxers

xoddam — Mon, 02 Feb 2009 01:24:59 +0000

We already know he wears socks with sandals. I for one choose to remain ignorant about the underwear preference of our glorious BDFL.

Did Linus Jump Too Soon? (Linux Journal)

smokeing — Sun, 01 Feb 2009 23:58:17 +0000

In Linus' shoes, if my chosen DE announced a push for such an extreme case of cross-platform integration as KDE's flirtation with Windows, I would at least feel queer.

He is known to be quite unexcited by all the virtualisation effort, and rightly so: why would one so intimately involved with Linux ever care to look at other platforms? And, when he sees a potentially enormous amount of work at KDE going to get it running, absurdly as it sounds, "natively" on Windows, why, he would so naturally turn away.

(On an unrelated note, in that same interview, he openly flounts the public opinion and all the journalist folk quipping thusly (http://www.computerworld.com/action/article.do?command=vi...): "Hey, I usually do my presentation slides in PowerPoint." Subtle and enlightening, isn't it?)

Weather applets

madscientist — Sun, 01 Feb 2009 22:10:10 +0000

He said "radar pictures". I don't know where you are but many areas provide relatively up-to-date radar, updated multiple times an hour; even every 5 minutes or so.

I too find this very useful (although I use Gnome and their weather applet); for New England winters getting out of work before the snow hits can make the difference between a 45 minute commute, and a 4.5 hour commute.

For whatever it's worth I find it more helpful as an applet, than looking it up on the web.

Did Linus Jump Too Soon? (Linux Journal)

dlang — Sun, 01 Feb 2009 21:32:57 +0000

what is it with bluetooth that you are seeing as non-functional? admittedly I don't do much with bluetooth, but I'm running KDE 4.x (on ubuntu 8.10) and it seems to be working for me.

there was an update to ubuntu in the last week that recognises the built-in bluetooth on my T61 for the first time, but I don't know if that is a kernel, ubuntu, or KDE update.

Did Linus Jump Too Soon? (Linux Journal)

dlang — Sun, 01 Feb 2009 21:14:22 +0000

where is it that you live that the weather prediction isn't valid for more than 15 min?

and what weather service is it that you use that updates itself that frequently?

everything that I've ever looked at does (at best) a day at a time with 'local' temperature readouts (usually one per city or less)

Did Linus Jump Too Soon? (Linux Journal)

truggieri — Sun, 01 Feb 2009 20:52:21 +0000

The only question really is which DE is Bloat and which is UltraBloat.

Did Linus Jump Too Soon? (Linux Journal)

eru — Sun, 01 Feb 2009 20:49:19 +0000

Personally, I consider telling me what the weather may be like in a few hours time to be one of the more useful things that my computer can do.

Precisely. But I find it is also one of the less useful things a desktop computer can do. I don't lug it (or even a laptop) to places where I might want to known the weather. But I can nowadays easily get a weather report on my mobile phone (which is really a pocket-size always-connected computer) practically everywhere I might feel the urge to look it up.

Did Linus Jump Too Soon? (Linux Journal)

callegar — Sun, 01 Feb 2009 18:55:01 +0000

In my opinion KDE 4.2 is generally reasonably fast. I am using it on an older Athlon PC with an old Radeon 9600 graphics board, and it is pretty usable on the speed front, even with effects on. What I see as bad is that the road from 4.0 to 4.2 still has not brought us bluetooth and that many many things seem not to have adequate docs (I spent one afternoon trying to understand why kded was dying at every login, just to discover that I had to erase the cache in /var/tmp to make it happy).

Did Linus Jump Too Soon? (Linux Journal)

sebas — Sun, 01 Feb 2009 18:47:30 +0000

The slowness you perceive might be caused by lacking video drivers. Especially NVidia had some issues here in the past, most of which are fixed with the 180.22 release. That's painting performance, and it might vary per video driver, unfortunately.
Application startup should be comparable or faster than in KDE3. If you experience problems there, make sure it's a local problem, if it's not, please let us know what exactly goes wrong -- we can fix it most likely.

Overall, from the experience of many users, KDE4 does feel snappy and fast. So it's most likely nothing general but an issue specific to your setup, meaning you don't have to accept slowness.

Honestly, if you call the speed of development slow, then I'm just puzzled by your assessment and wonder what you compare it with.

Did Linus Jump Too Soon? (Linux Journal)

sebas — Sun, 01 Feb 2009 18:40:08 +0000

Yeah, so give the fine weather applet that just entered KDE extragear a shot. If you need multiple places, you can add more than one to your desktop. If you'd like a more basic weather applet, choose weather station.

For other critical features, there's also a Plasma version of XEyes.

Did Linus Jump Too Soon? (Linux Journal)

bronson — Sun, 01 Feb 2009 17:17:24 +0000

This is a meaningless statement since it applies to pretty much every application and feature I can think of: "I need email" "Visit gmail" "I need an IDE" "Visit Heroku" etc. I guess there's no need to spend any time developing DEs anymore!

Also, 15 seconds adds up to a lot of time when you're checking many times each day.

Did Linus Jump Too Soon? (Linux Journal)

robert_s — Sun, 01 Feb 2009 16:40:25 +0000

So visit a webpage. It's 15 seconds more work, but it shouldn't be something that makes a desktop unusable.

Did Linus Jump Too Soon? (Linux Journal)

endecotp — Sun, 01 Feb 2009 16:35:29 +0000

> most would regard [a weather applet] as "useless eye candy".

Presumably those people live in places where the weather is predictable. Personally, I consider telling me what the weather may be like in a few hours time to be one of the more useful things that my computer can do.