Slashdot Mirror
Linux, to be (Like Microsoft) or Not to be?
David writes "Stephen Shipman delivers a very articulate and concise view of how Linux fits in server and end user environments. He expresses his view in response to Nicolas Petreley's 'rant' in Linux Journal. He points out the subtle implications of efficiency versus consistency." From the article: "[...] efficiency (as measured by keystrokes) isn't the only metric for ease of use. Consistency must also be taken into account. Microsoft has made a lot of hay (and green) by flogging consistency".
case in point: the pop-up key-stealer, when apps suddenly thrust themselves forward and steal a keystroke for the [ignore] [retry] [cancel] [OK] whatever prompt and vanish if it meets the input expectation.
The what? I vaguely recall that being a problem in Win98, but I run Win2k here, and when an inactive application demands input, it stays right down in the taskbar where it belongs - all that happens is that the taskbar icon flashes to notify me. Surely this is the case in WinXP too? It would seem strange for Microsoft to introduce the correct behaviour in one version of Windows, only to take it out again in the next.
What I repeatedly hear from Mac enthusiasts is how quickly a new user can sit down and get right to business, without thinking half as hard where things are or how settings work.
And that's total bullshit. OS X is arguably easier to learn for someone who's new to computers altogether, but anyone who has only ever used Windows before, faced with a Mac, is going to have a terribly frustrating time just trying to resize a window ("I click on the left edge and drag, to make it wider, and the window moves instead! What's with that?"), let alone figuring out how on earth the Dock is supposed to work.
What it comes down to is, people like what they're used to. That means Mac users love Macs, Windows users say they hate Windows but hate trying other platforms even more, and Linux users can't figure out how anyone can find Linux difficult to use. Which is why it is sensible for Linux to behave more like Windows (KDE), or more like OS X (Gnome) - because with greater familiarity will come greater uptake.
All OS's have little backwards bits here and there. Gnome has gconf, for instance. MacOS X has some hidden config files you have to get to for (rare) things.
Compare, say, setting up apache on a typical Linux distribution with configuring IIS on Windows. The difference is night and day. Sure, sometimes you have to dig into the Machine.conf or use a command line tool like httpcfg, but these are rarities, rather than the common case. Also, while there are some GUI configuration tools for apache from various sources, all of them suck rocks through a straw to the point that it's EASIER to look up arcane flags and configuration settings and type them into a text editor than it is to click a button. Typically, it's just a graphical representation of the config file.
OSX and Windows do a damn good job of making the common stuff easy to configure and use with a nice GUI. On Linux, what config applet you use may depend on which environment you're using. KDE and Gnome both have different stuff, as does SUSE, Red Hat, etc.. consistency may be better (not great, but better) within one particular distro, but not across even two similar ones.
This is a hot button, though. Lots of people will disagree, because whatever they're doing works for them. it's that kind of myopic outlook ("it works for me, you must be too stupid") that makes it so difficult for Linux to gain acceptance. It doesn't have to "work for you", it has to "work for THEM", and if it works for you too, then that's even better.
If you need web hosting, you could do worse than here
No matter how they pass XP off as a multi-user environment, it carriest considerable baggage of being single user - case in point: the pop-up key-stealer, when apps suddenly thrust themselves forward and steal a keystroke for the [ignore] [retry] [cancel] [OK] whatever prompt and vanish if it meets the input expectation.
Of all the things you could propose as a reason for considering it "single user", that's the oddest. It's hateful and frustrating, and more prevalent in MS WIndows than X11 or Mac OS, but it's more prevalent in X11 than Mac OS, and more prevalent in Mac OS than 8 1/2.
You could have pointed to the single-application-instance shared with Mac OS (which Firefox has imported to X11). Whether it's services, desktop applications, or just logged in users, it takes a huge effort to have two instances of ANYTHING running in Windows.
Their virtual terminal and user switching required years of development work from Citrix, Xerox, Metaframe, and other companies to figure out what parts of the user environment should be shared, what should be duplicated, and what should be switched from instance to instance... and you still can't have two login sessions under the same user id.
For applications that run as services there's been even less work done to get around the problems... so it's actually more cost effective to build "blade" servers or run multiple copies of the OS in virtual machines than to run multiple webservers or other applications in the same instance of Windows.
I mean, I had a 486/50... this is a machine that wasn't powerful enough to run one instance of even NT 3.51... and I was running multiple webservers on different addresses under the same kernel. This kind of thing is routine and easy in UNIX, because it was designed for multiple users (and thus multiple instances of every possible resource) from the very start.
It has, as the logical postivists say, "no cognitive content", or at least very little. By talking about "Linux" and indeed "Windows" so broadly, you can make the figures for consistency come out to whatever you want. In either case the largest source of inconsistency is the choice of optional software you choose to put on the system; as it is much more convenient and you have a much wider variety of software you can install on a distro like Ubuntu, naturally you can easily make your system wildly inconsistent. It's because there's so much software, from different sources, that are available at a touch of a button under Linux. A lot of that software is of course really bad from a UI perspective, but even if you restricted yourself to reasonably good software, it's still easy to end up with a LOT of software installed on a Linux box.
None of which of course applies in the server domains, where you're better off with less UI. Wildly divergent configuration files are bad, but not as bad as wildly divergent GUIs.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
For example, plenty of Windows users will be quite happy by going to "file" to print or close an application. "Find" is under edit, not view. That's fine for people who think that way and for them it SHOULD be that way. The rest of the userbase shouldn't have to suffer for it, though.
Myself, I like visuals. The idea of dragging an application window to a printer, OR dragging the printer to the application windows, appeals to me. (To me, drag&drop needs to work by object, not by destination.)
"But writing all those interfaces would be massively overwhelming!" I'm not suggesting anyone does. Just provide a rational, consistant, standard skin that the majority can use, then provide a powerful enough engine that can handle application look&feel and drag&drop events not otherwise handled. Then write a simple UI editing engine. If people want their own UI, give them the tools to provide it.
"Most people wouldn't bother." Probably true, but the Open Source dictum is that some will, and that evolution will lead to superior interfaces.
"How does that benefit company X that sells products?" Easy enough. Every time you're about to release a next major version, look and see whether other skins are doing better than your default. If they are, switch. If that's how everyone sees your program anyway, it won't hurt anyone's ability to use it.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
That's programmable through the window manager, though. With my current setup of Gnome, if I launch a new program, it pops up in the background, rather than in front.
Best UI improvement I've ever seen in the computer industry. I can start something, then keep working away on whatever I was doing before I started it, and when I'm ready, the program I started is up and running behind whatever I'm currently working on.
Having said that, it's not so simple as "everything starts in the background". It depends on whether the program has any open windows already, and what layer they're at, whether the program was started by another program or the Gnome menu, and a whole bunch of other crap. The way it's done though, seems to be very good.
"City hall" in German is "Rathaus" Kinda explains a few things......
I am not a windows user, but I happen to admin a mixed network with 50%+ Windows. With all due respect you are talking BS. This was valid in the days of 3.11. It has not been true ever since. Less then 5% of the applications nowdays will operate correctly if installed by copying because they rely on registry settings put in by the installer.
Funnily enough the model you are describing works fine on guess what... Gentoo and BSDs. Portage. I personally dislike it, but that is a matter of taste.
Baker's Law: Misery no longer loves company. Nowadays it insists on it
http://www.sigsegv.cx/
Linux: An operating system kernel.
Microsoft: A multinational corporation.
Unless the laws of reality turn in on themselves, I do not think Linux is going to become anything even remotely like Microsoft.
Linux got to where it is today by being both better and different from Windows, not by trying to be a cut-rate knock-off.
To play devil's advocate - Linux did get to where it is today by being a cut-rate knock-off. But it was a cut-rate Free knock-off, and it was a knock-off of UNIX, not Windows.
Linux has since surpassed many competitors in many ways, and has killer features that no longer relegate it to being a "cut-rate knock-off", but that's what it grew into when it became more than a hobby, and that's what enabled it to become as popular as it did in the mid-to-late 90s.
And your description here indicates that you understand neither the Windows nor the *nix way of doing things.
Windows: What if one of those windows applications also installs a service that's running on your box? What's going to happen when you delete "Program Files\App Name"? What about all those registry entries? What about the entry in "Add/Remove Programs"? What about dependencies? To me, this is why users shouldn't be deleting program files, and part of the reason why windows asks, "Hey, this is an executable. You really, really, REEEEEAAAAALLLY sure you want to delete it?"
Unzipping AppName.zip into a new directory in Program Files, then later removing the whole directory is quick, easy and clean.
I've never seen this done. Ever. And I wouldn't want it this way. I kinda like having stuff accessible in my Start menu. This is why installers exist - the presence of a piece of software on a system goes beyond just the executable(s) and data files. Software has to be installed into the target system. An engine that sits in the bed of a pickup truck isn't installed, and isn't going to be very useful for moving the truck around.
Linux: Did you install from source? If so, did the source installation come from a ports tree or similar? If so, it created an entry in the package database, which keeps track of how to uninstall the software. If you installed bare source outside of a ports tree, then you're responsible for either:
- Keeping the source code around, so you can do a 'make uninstall'
- Keeping track of what files were added/removed/modified manually.
In either case, you're a proficient enough user to understand how to build from source, you should be able to understand how to administer a system with software built-from-source. Talking about binary packages after all this is a moot point - you use the package manager to uninstall the package. Why would you uninstall something using a different tool than you used when you installed it?or
As a for-instance, my desktop system has approximately 380 packages installed. I don't want to scour the unix manpages, makefiles, or anything else (e.g. if I was running windows, the registry) to figure out how to uninstall a program, even though I'm a comptetent, experienced system administrator on Windows, Linux, and a few Unixes. The package management tools on Windows and *nix make it possible to feasibly maintain systems with hundreds (or even thousands, if need be) of software packages installed. In your example, you think uninstalling means deleting the executables and data files. In both cases, you're missing the point - uninstalling an application should be done through an appropriate tool. When you ask the question, "How do I uninstall a program?" for a common user, the answer always is:
- For windows, it's control panel\add/remove programs
- For *nix, it's the standard system administration tool you've always used for everything else.
Don't go mucking around deleting files. This isn't a Linux-only idea.Confusing and frustrating in my opinion.
What's confusing/frustrating about double-clicking on a package (.exe,
I pity the foo that isn't metasyntactic
Vista is pretty much multi-user on the Unix level
No it's not. MS will never make a multi user OS. MS wants one user only per copy of windows. They don't like multi user systems. They don't like thin client systems either. (yes, MS sells a terminal server. They had to get into that cause some businesses wanted to go that route. But they're still doing their best to implant the single user mentality to average users. Their mainstream OS has no such capabilities. Plus, MS terminal server is a joke as you pay both for the server and per user. So all cost benefits are negated)
God forbid if people find out what multiuser systems really do. Imagine that! Joe Sixpack would start wondering why he needs a copy of windows for each member of his family. Or maybe he'd go wild and just use dumb terminals to connect to the main computer. (for the uninitiated: dumb terminal doesn't mean 'command line'- you can have your desktop as you know it, running super fast on a cheap and light machine, like a pII-200MHz 32mb)
MS threw dust in your eyes, pretending they have a multiuser OS, just so they don't get behind in the catchwords race. Before you say "but i can have more than one user on my windows machine", can multiple users use the pc at the same time? Try it. Login to your account, then create another user on that machine. Now try to login remotelly with the new user, while you're still logged in with your previous username. Windows will throw a popup, warning you, that another user is logged in and if you continue that user will be logged out. There can only be one!
And ofcourse, even if MS changes tomorrow and decides to go multi user, there is a single-user culture in windows, that's difficult, if not impossible to change. ie. most windows apps would not work.
VStrider.