Exploring Linux Desktop Myths
Krafty Koder writes "Over at Newsforge (Part of OSTG, Slashdot's Parent) there's an interesting article that attempts to dispells the myth that Linux isn't ready for the desktop or that Windows still beats Linux.
Three myths are explored - that Linux is harder to use, difficult to install and that there's not enough apps ."
Your so incredibly right. Linux is in desperet need of something as simple as the windows install sheild so you dont have to read a manual to install anything. Other then that, I think linux is leaps and bounds head of what joe average gives it credit for.
You mean just like you don't have to on Linux, either? Thanks for re-affirming the fact that difficult application installation on Linux is a myth.
Funny you mention that. InstallShield is available for vendors to use if they want to create an installer for Linux.
Why the heck vendors are sticking with their own crappy command-line-only installs is really beyond me. Only one I know of that uses IS for Linux is if you download the Java NetBeans Cobundle.
Actually, UnrealTournament 2004 may also use ISX. I remember some nifty Java installer....
--
Given enough personal experience, all stereotypes are shallow.
I think the issue here is you're more likely to have to use make to install FREE software on linux. Just as you would use it to install FREE software on the Mac. If you're BUYING software for linux, you can bet you'll get an installer or an rpm or something. That's one of the things I don't see mentioned a lot is that we're comparing the ease of install of a corporation's proprietary software to the ease of install of in-progress open source stuff. Most things that you would pay for otherwise, such as office suites, web browsers, etc. do have installers (Open Office, Mozilla, etc). The things that you use make to install you probably don't even have access to on an M$ system. It's like complaining that Dell sent you a free graphics card but you ahve to install it yourself when you could have paid Best Buy $200 to install it. It's not quite fair to compare the M$ office installer to a network sniffer program you found for free on some h4x0r d00d's site. It seems this happens a lot.
My experience with Windows is that, so long as nothing goes wrong, there isn't much maintenance. But, when problems arise, even finding documentation on how to fix things is a problem, despite its popularity.
Recently, I needed to change the subnet on the public side of a group of computers, including a Win2K server and some Linux boxes. The Linux boxes took 5 minutes each, and all the daemons adjusted to the changes with a simple 'service XXX restart'. Never even had to reset the hardware.
It also took 5 minutes to change the IP on the Win2K box... but it required a restart... and several very important things didn't come back afterwards (Exchange and RAS in particular). It took two days to track down WHY, because everything looked correct.
Turned out that the settings were correct, but neither Exchange nor RAS are tolerant of changes to the binding order of interfaces, and any significant change (like an IP) to an interface changes its binding order... making it invisible to both applications. This isn't documented in anyplace convenient; I only found it by tracking down specific error message text via GOOGLE, piecing together information off of several of the results, finally finding a very nice MSKB article on the subject... which hadn't appeared in the GOOGLE search results!
I'm now fighting a problem where this same machine has decided that its second and third net cards are deaf... They exist, they detect the network, Win2K says they're working, but the rest of the network can't talk to them. And the customer gets frustrated, while I try to find SOMETHING on the net that would explain the behaviour. I could have fixed this problem within an hour under Linux...
As Dogbert said, "We have the very best kind of evidence. Anecdotal!
A large variety of useful applications are simply not available on things like the red hat network (so there goes up2date).
Up2date is great for upgrading packages that I already have installed during the first little while after I buy the operating system (at the moment, I think I have 3 years, but only because I bought "Enterprise Linux" which costs more than MS Windows, so there goes the cheaper concept.
Up2date will get me new software if I happen to know what the package name that I want is. For instance I wanted to install GVIM. The redhat network package for Enterprise Linux was compiled without support for the GUI, alas. This is what I had to do to get GVIM.
I suppose I could have gone to the VIM website, and download their version of the vim RPMs, and ran rpm --install on that. It probably even would have worked, but I don't know for sure that vim.org has the same RPM version, if it installs everything in the same place, and what up2date would think about that package. I have only been using Linux for approximately 2 years at this point, so it is perhaps understandable that a complete newbie like myself would find the various methods of software installation extremely complex.
Next topic: configuration
My biggest complaint against linux on the desktop is the extreme absurdity of some types of configuration. My linux workstation is my work machine, and so it has to do things like run apache and our application server (Interchange if you are curious -- http://www.icdevgroup.org).
Well... Installing apache was easy. Installing Interchange was hard. It requires nonthreaded perl, but a gazillion things require the existing threaded perl 5.8.3 that came with the OS. So, I build my own perl from source, and then copied into /usr/local instead of /usr, and then hacked at the Interchange source so that it looked in this new location...
Why not just uninstall the old perl? Because I can't. When I issued the
command, I encountered dozens of errors. And I couldn't upgrade -- they were the same version, just with different compiler switches set.
Oh, then I got an error that my system was set to the incorrect language encoding. I'm not 100% sure I understand what that means even having fixed it -- but here's what I had to do
Ok... so... exactly what does i18n stand for? Having a normal computer user find the i18n configuration file and hack at it isn't reasonable. Especially if you don't tell them that the error is in i18n. Now a normal user might not be trying to install Interchange, but hell, even our sysadmin wasted 4 hours figuring out what config file to change.
My point? Linux on the desktop is a freaking toy until a user can do everything they need to without opening a terminal or becoming root ever.
This includes installing software, configuring the machine, and running applications.
Incidentally, Windows crashes since I installed XP (2 years): at least 10, at most 15.
Linux crashes since I installed Red Hat Linux (first v. 7.2, now EL 3 WS, approx 18 months): at least 10, at most 15.
My definition of a crash is anything that goes wrong to which the only easy solution is either pressing the power button or typing "init 6".
No it doesn't have any problems displaying open/close form tags. If IE is not displaying the extra new lines then it's actually doing it wrong. Form tags, as specified in the w3 standard are block-level elements.
Block-level elements are those elements of the document language that, by default, are formatted visually as blocks (e.g., paragraphs). Inline elements are those elements of the document language that do not cause paragraph breaks (e.g., pieces of text, inline images, etc.).
That is why you get the extra space around the element. This is the intended behavior of form elements. If you want to get rid of the space, as mentioned, use a style sheet of margin:0; or display: inline;
[alk]