Issues include. Difficulty installing software and hardware. Having to RTFM to do anything. Difficulty in viewing common formats like PDF (No, block characters and unreadable text is not sufficient even if the file does open). The GUI is still early 90s feel at best.
My immediate reaction is, wow, what a crummy distribution.
x86-64 Fedora Core 4 literally "just works" for all those things, and the only config files I had to hack was/etc/fstab (because I have an extremely wierd disk layout -- but I'm sure there's a GUI tool I could have used). With KDE 3.5 it looks and performs beautifully, and beats Windows XP hollow for absolutely everything (except Half-Life 2: the only reason I still have a Win32 partition on this thing).
Installing it was pretty much a case of clicking "Next" repeatedly and then wandering off for half an hour while it downloaded and installed packages. Believe me, it was far less difficult than the last time I reinstalled XP, a multi-hour battle involving many reboots, wrestling with Windows update and searching for drivers for hardware that worked out-of-the-box with FC4.
I keep hearing all these horror stories about how difficult Linux is to install and use, to my utter bemusement.
OpenOffice itself is very basic and cluttered; It probably doesn't have a need to use a format that supports more than the basic TEXT GOES HERE system.
Please wait while I wet myself laughing at your ignorance.
From your post I have extracted the following metadata:
You installed an early (1.1.x) version of OpenOffice, looked at it once or twice (without actually doing anything with it), and then decided you didn't like it because it didn't look exactly the same as MS Office
You have absolutely no technical knowledge of the ODF format whatsoever
Believe me, describing the ODF format as a "basic TEXT GOES HERE system" is like describing an immersive visualisation facility as a "computer display".
In particular, a US F-117 Stealth fighter was shot down over Bosnia. The shooters could not track the plane on radar -because it's stealth, you know- so they looked instead at the changing signal patterns of the cell system as the plane flew over.
Not entirely true. From Wikipedia:
According to Wesley Clark and other NATO generals, Yugoslav air defences tracked F-117s with old Russian radars operating on long wavelengths. This, combined with the loss of stealth when the jets got wet or opened their bomb bays, made them highly visible on radar screens. The pilot survived and was later rescued by NATO forces.
The maths has also been done to show that you can use emissions from FM radio station transmitters and a pair of receivers (with a baseline of about a mile) to track F-117s easily (as long as you have enough computer power). This is one of the reasons that one of the first bombing targets in the Iraq war were civilian radio station transmitters.
The B2 doesn't suffer from this vulnerability -- it doesn't rely on geometry so much as materials that don't reflect radar radiation.
Yes, so the X server is running on the thin clients, and the X clients, i.e. the applications, are running on the server. So there should never be any need to run an X server on a terminal services server.
X is confusing because the client/server are the "wrong way round".
And killing someone silently hasn't changed one bit as preferably over shooting him with a sniper rifle (which you can not silence, no matter what Hollywood tries to tell you).
Depends on the calibre. You can kill someone very effectively with subsonic.22 rimfire ammunition, if you can get a good shot. And.22 hunting rifles are fairly inexpensive, and quite easy to get hold of. I've fired one -- it came with a suppressor as standard, and the only noise it made was a 'ping' as the hammer struck the firing pin.
Anyway, if you choose your spot carefully, you could easily get away with using a full-power 7.62 or even.338 calibre weapon. Don't underestimate the disorientating effect of a supersonic round passing a couple of feet away. By the time the security detail have stopped shitting themselves, you can be well on your way from your perch (anywhere within 600 yds will do, usually, and that's a hell of a lot of ground to search in anything other than a flat, barren wilderness). The problem's not the equipment, the skills or the risk of getting caught: the problem's having steely-cold will enough to look at someone's face through an optical sight, knowing that if you pull the trigger, that person's life will end.
And you wonder why the SS are so paranoid about assassination attempts?
What do you XOR against? Should I assume 0xFF Or did you fuck up and mean "Double ROT13"?
It doesn't matter what he XORs against, as long has he XORs against the same thing twice. I think you should go and beat yourself over the head with a clue stick. Repeatedly.
Generally it's a good idea to do some basic fact checking before you start mocking someone.
Why? Maybe it's ok for engineer types. Programming is a Mathematical activity so why torture your self, the best language to start is Lisp/Scheme.
Because computer scientists don't design integrated circuits, engineers do. So learning how processors work at the fundamental level is quite a large component of our information engineering course this year.
At the other end of the scale, our mathematics course this year contains MATLAB & C++ software engineering[1]. But by working at both ends of the spectrum, I think we get a better view of how programming works.
[1] Notably, the emphasis in the C++ course is on the importance of starting out with a good design on paper before writing any code... AFAIK, we don't do any pointer arithmetic or object oriented stuff until the specialized computing courses in 3rd year. I personally think Python would have been a better language to start us off with...
All fresher Engineers here (Cambridge, UK) have to learn to program 8086 assembler. Except they don't get an assembler, they have to enter the program using a hex keypad on the front panel.
Yes, in 2006. And no, I'm not joking, I did it last term.
I've always liked saving my resume out in rtf format when companies asked for.docs. I guess I had some idealistic notion that I was resisting MS somehow, but, as I knew even then, I was counting on rtf being mapped to Word if they had it installed on their box. One day to check on the xplat of rtf, I took a look at my rtf resume on some alternative word processors on my Mac. Long story short, an *.rtf file looks quite a bit differently when swapped between Word, AppleWorks, TextEdit, and AbiWord.
I'm sorry, I must have missed something in your post I think; what was the problem with sending them your resume as a PDF file?
Seriously though, I would really appreaceate it if someone came up with a way to standardise installing and updating programs in Linux -and make it so that it only requires a few clics.
This is where I am suddenly confused. It's called a package manager! It's supplied with your distribution! It gives you access to thousands of programs, all of which can be installed (with all their dependencies) using only a couple of clicks! The latest version of 'yumex', the graphical frontend to the Fedora Core package manager, will let me update all of the software on my computer (excepting the very obscure) in only a couple of clicks.
However, I do see vaguely what you're getting at -- you want the installation process for any program to be something like the way programs are installed on Windows. I'll spare you the "technical mumbo-jumbo", and just say that there are technical considerations that preclude that that are too numerous to count.
If you wish to continue this interesting discussion, please feel free to e-mail me!
What makes you think they wouldn't install the same amount of crap on the Linux machines? Out of curiousity...
I've no doubt they would. However, most of the equivalent software on Linux starts at boot time rather than at login time, and runs as a daemon that stays running between user sessions, rather than being (re)started for each session. The main bottleneck on the current systems is the hard disk (CPU load and RAM usage seem fairly reasonable); if the crap was already memory-resident when I came to login, I don't think I would notice it at all.
You literally would not believe the amount of crap they've put on these systems, and they've very carefully locked it down so that you can't get rid of it. Amusingly, the extent to which they've locked the systems down renders several of the programs they've installed totally useless!
Note to self: write bitchy e-mail to the computing office complaining about the login times....
Another problem I think, is when you comment on some rather basic problem with regards to Linux, you usually get alot of technical mumbojumbo as a reply. I just want to install a damn program, it can't be that hard can it?!?
Well, you are posting to Slashdot, so I assumed you wanted an actual answer. I could have said, "That's just the way it is," but that wouldn't have been helpful. As far as just installing "a damn program"... that's very dependent on which program it is, and believe me, that's not an issue peculiar to free software.
I'm curious; which particular programs did you want to install that weren't available from your distribution's package repository? And what particular difficulties did you encounter? It's hard to fix problems without specifics, after all.
As I said, it's to get from a login screen to a usable desktop... the computer services office here install so much crap on the XP systems that it actually beggers belief.
Having been a long-time W2K user before switching to Linux, I can tell you that W2K would boot on a 500 MHz Celeron laptop faster than these dual P4 systems log me in.
Universities don't tend to have bleeding edge workstations.
My college at Cambridge just got some new desktops. They're dual-P4 systems with a gig of RAM each, and they're primarily used for word-processing and surfing the net. Guess how long they take to go from Windows' login screen to a usable desktop?
FIVE GOD-DAMN MINUTES?!?!?!?!??!!
*beats head against desk*
This is while the Engineering department use Knoppix (well, MDP) workstations that take 5-10 seconds to log on to and are just as usable....
You make some valid points. Unfortunately, things like the lack of out-of-the-box MP3 support and the problems talking to XP systems are far more "can't fix" than "won't fix." There's literally nothing that free operating system developers can do about patents and moving-target closed specifications. And it's really not the fault of the people who write free software that most mainstream game developers seem to have no interest in supporting their platforms. On the otherhand, you say you tried Fedora Core. To make MP3 support, DVD playing, etc available to the package manager takes one command. Got to http://rpm.livna.org/ and see how long it takes you to find it... (and before you say Livna is obscure, it's easily found by Googling).
However, in saying that you want click-and-go install of applications without compiling, you're missing the point. This is only possible using a distribution's package management system, or distribution-specific click-and-go systems such as Klik. Why? The reason is that precompiled software is architecture- and platform-specific. A program compiled to run on x86 Slackware will never work on PPC Gentoo, and vice versa. In addition, consider that software seems to be written nowadays to run on either Windows, or anything except Windows. It will never be possible for developers to provide binaries for every possible combination of kernel+libraries, so they spend vast amounts of time writing e.g. autoconf scripts so that (hopefully) all you have to do is type "configure && make install" -- a process I don't consider to be too difficult to understand.[1][2]
I would take issue with your statement that working with Linux is like "[Figuring] out how the damn carborator (sic) works when I mainly want to drive." If you want to fiddle with the innards of Windows, you have to drill down into obscure, poorly documented configuration dialogs, and brave the mess that is the registry (why the hell should I have to enter binary data into an undocumented registry key in order to do something as simple as swap my Ctrl and Caps Lock keys?) If you want to fiddle with the innards of Linux, most things are configurable from reasonably-named plain-text configuration file somewhere on your system. Having said that, the last time I had to do fire up an editor and hack a config file was when I was trying to do something decidedly non-standard... several months ago.
I'd say that moving to Linux from Windows is like moving from driving a family car to driving an aircraft; although the controls are very different, once you've mastered them you get where you're going much more quickly and comfortably.
[1] Yes, I realize that bytecode-compiled languages and interpreted languages allow one binary to run on many systems. But they're mostly relatively slow.
[2] I agree that when "./configure" fails, it's often quite difficult to work out what's gone wrong. Many developers aren't good at writing meaningful error messages for the self-configuration stage, unfortunately.
Yes, and the problem with Linux is not the lack of applications, rather, the fact that installing any is a total bitch.
That is not true, to a ridiculous extent. I use my Linux computer for some very obscure things: VHDL coding and simulation, CAD and low-level serial IO, some of which I do while logged in remotely. All of the tools I required for that were available to install using a simple graphical utility that took care of all the dependencies for me.
Lets add the fact that I had to reinstall my computer from scratch a month ago, and it took me:
Windows partition: some games, a browser, a media player and a IM client along with drivers took a day to wrestle into submission. Not to mention I had to reconfigure my personal settings from scratch, which took about the same time again.
Linux partition: many obscure and complex development tools, office applications, media players & lots and lots of other stuff: 5 hours, most of which was spent away from my computer while the package manager did its stuff.
You, sir, are a troll, who clearly hasn't tried a Linux distribution released since 2002.
My department looks for someone who can sit down and do a board design with 2 GHz digital ECL chips mated with RF components, and there's just nobody.
Guess why that is? You don't learn how to do that even at university. And no company will give you a job doing that if you don't already know how to do it, because they don't want to have to pay you while you learn. Not to mention the fact that it's impossible to do in your own time because the layout and simulation tools needed to create a working design are ridiculously expensive.
The thing that NEEDS automation is creating schematic symbols & footprints. For real now, there's just no excuse for that.
The problem is that creating good symbols is a bit of a black art, in choosing how to lay out the pins and whereabouts on the symbol to put the buses. I don't see anyone automating it in a sensible way any time soon, unfortunately.
PCB now comes with a whole bunch of parameterized footprints now, though, which are quite nice. Also, a number of footprint-generating scripts for various types of footprints are out there -- you just have to look for them. However, I haven't yet found a footprint that _wasn't_ already available in PCB...
Linux is ok for openoffice, internet, mail and programming, but if you actually want to *do* something with the computer that isn't programming there's no wealth of professional grade software out there for professionals who are reliant on computer technology.
There's a lot of professional grade software out there for professionals that runs as well or better on Linux than on any other operating system:
Pro Engineer
Various Cadence microcircuit design toolkits (schematic capture, mask layout, simulation)
Xilinx FPGA design & synthesis tools
I could carry this list on for a long time, and that's just in my field. I keep on having to repeat myself: for high-end engineering research and development, people are upgrading to Linux all the time.
Cambridge University Engineering Department use Linux workstations for all undergraduate tuition -- they're based on a custom Knoppix distribution. More information at the CUED website.
Professional software for professionals: lots of it for Linux. Maybe you mean "professional" software for amateurs?
We have FPGA development boards in our lab that are specifically designed for signal processing.
Yes, but my suspicion is that they consist of high-performance ADC at one end, a DAC at the other, an FFT, and an IFFT, with some digital logic tying them all together.
Most cutting edge signal processing is done in the digital domain: for instance, using FFTs it's possible to build a 100 kHz high-pass filter that has a stupidly high Q-factor.
If a government is asking for backdoor access to one operating system, why not ask for backdoor access to others? How would the open source community handle such a request when a government comes insisting on a backdoor to your favorite flavor of Linux, or OSX, or BSD, or something that hasn't yet been developed?
Give them the finger, of course. Then post the e-mail to Slashdot.:P
Slashdot Mirror
My immediate reaction is, wow, what a crummy distribution.
x86-64 Fedora Core 4 literally "just works" for all those things, and the only config files I had to hack was /etc/fstab (because I have an extremely wierd disk layout -- but I'm sure there's a GUI tool I could have used). With KDE 3.5 it looks and performs beautifully, and beats Windows XP hollow for absolutely everything (except Half-Life 2: the only reason I still have a Win32 partition on this thing).
Installing it was pretty much a case of clicking "Next" repeatedly and then wandering off for half an hour while it downloaded and installed packages. Believe me, it was far less difficult than the last time I reinstalled XP, a multi-hour battle involving many reboots, wrestling with Windows update and searching for drivers for hardware that worked out-of-the-box with FC4.
I keep hearing all these horror stories about how difficult Linux is to install and use, to my utter bemusement.
There's a distinction between append and insert?
Dang. *goes back to sweet, sweet emacs nirvana*
Please wait while I wet myself laughing at your ignorance.
From your post I have extracted the following metadata:
Believe me, describing the ODF format as a "basic TEXT GOES HERE system" is like describing an immersive visualisation facility as a "computer display".
</flame>
Not entirely true. From Wikipedia:
The maths has also been done to show that you can use emissions from FM radio station transmitters and a pair of receivers (with a baseline of about a mile) to track F-117s easily (as long as you have enough computer power). This is one of the reasons that one of the first bombing targets in the Iraq war were civilian radio station transmitters.
The B2 doesn't suffer from this vulnerability -- it doesn't rely on geometry so much as materials that don't reflect radar radiation.
Yes, so the X server is running on the thin clients, and the X clients, i.e. the applications, are running on the server. So there should never be any need to run an X server on a terminal services server.
X is confusing because the client/server are the "wrong way round".
Depends on the calibre. You can kill someone very effectively with subsonic .22 rimfire ammunition, if you can get a good shot. And .22 hunting rifles are fairly inexpensive, and quite easy to get hold of. I've fired one -- it came with a suppressor as standard, and the only noise it made was a 'ping' as the hammer struck the firing pin.
Anyway, if you choose your spot carefully, you could easily get away with using a full-power 7.62 or even .338 calibre weapon. Don't underestimate the disorientating effect of a supersonic round passing a couple of feet away. By the time the security detail have stopped shitting themselves, you can be well on your way from your perch (anywhere within 600 yds will do, usually, and that's a hell of a lot of ground to search in anything other than a flat, barren wilderness). The problem's not the equipment, the skills or the risk of getting caught: the problem's having steely-cold will enough to look at someone's face through an optical sight, knowing that if you pull the trigger, that person's life will end.
And you wonder why the SS are so paranoid about assassination attempts?
It doesn't matter what he XORs against, as long has he XORs against the same thing twice. I think you should go and beat yourself over the head with a clue stick. Repeatedly.
Generally it's a good idea to do some basic fact checking before you start mocking someone.
Because computer scientists don't design integrated circuits, engineers do. So learning how processors work at the fundamental level is quite a large component of our information engineering course this year.
At the other end of the scale, our mathematics course this year contains MATLAB & C++ software engineering[1]. But by working at both ends of the spectrum, I think we get a better view of how programming works.
[1] Notably, the emphasis in the C++ course is on the importance of starting out with a good design on paper before writing any code... AFAIK, we don't do any pointer arithmetic or object oriented stuff until the specialized computing courses in 3rd year. I personally think Python would have been a better language to start us off with...
All fresher Engineers here (Cambridge, UK) have to learn to program 8086 assembler. Except they don't get an assembler, they have to enter the program using a hex keypad on the front panel.
Yes, in 2006. And no, I'm not joking, I did it last term.
I'm sorry, I must have missed something in your post I think; what was the problem with sending them your resume as a PDF file?
Insert fancyheader LaTeX code into Layout->Document->Preamble.
Yes, I know you have to know LaTeX to do that, but you do it once, make a template out of it and then forget about it.
There are some good examples on the LyX site, but I don't have time to hunt them down right now.
This is where I am suddenly confused. It's called a package manager! It's supplied with your distribution! It gives you access to thousands of programs, all of which can be installed (with all their dependencies) using only a couple of clicks! The latest version of 'yumex', the graphical frontend to the Fedora Core package manager, will let me update all of the software on my computer (excepting the very obscure) in only a couple of clicks.
However, I do see vaguely what you're getting at -- you want the installation process for any program to be something like the way programs are installed on Windows. I'll spare you the "technical mumbo-jumbo", and just say that there are technical considerations that preclude that that are too numerous to count.
If you wish to continue this interesting discussion, please feel free to e-mail me!
I've no doubt they would. However, most of the equivalent software on Linux starts at boot time rather than at login time, and runs as a daemon that stays running between user sessions, rather than being (re)started for each session. The main bottleneck on the current systems is the hard disk (CPU load and RAM usage seem fairly reasonable); if the crap was already memory-resident when I came to login, I don't think I would notice it at all.
You literally would not believe the amount of crap they've put on these systems, and they've very carefully locked it down so that you can't get rid of it. Amusingly, the extent to which they've locked the systems down renders several of the programs they've installed totally useless!
Note to self: write bitchy e-mail to the computing office complaining about the login times....
Well, you are posting to Slashdot, so I assumed you wanted an actual answer. I could have said, "That's just the way it is," but that wouldn't have been helpful. As far as just installing "a damn program"... that's very dependent on which program it is, and believe me, that's not an issue peculiar to free software.
I'm curious; which particular programs did you want to install that weren't available from your distribution's package repository? And what particular difficulties did you encounter? It's hard to fix problems without specifics, after all.
As I said, it's to get from a login screen to a usable desktop... the computer services office here install so much crap on the XP systems that it actually beggers belief.
Having been a long-time W2K user before switching to Linux, I can tell you that W2K would boot on a 500 MHz Celeron laptop faster than these dual P4 systems log me in.
My college at Cambridge just got some new desktops. They're dual-P4 systems with a gig of RAM each, and they're primarily used for word-processing and surfing the net. Guess how long they take to go from Windows' login screen to a usable desktop?
FIVE GOD-DAMN MINUTES?!?!?!?!??!!
*beats head against desk*
This is while the Engineering department use Knoppix (well, MDP) workstations that take 5-10 seconds to log on to and are just as usable....
You make some valid points. Unfortunately, things like the lack of out-of-the-box MP3 support and the problems talking to XP systems are far more "can't fix" than "won't fix." There's literally nothing that free operating system developers can do about patents and moving-target closed specifications. And it's really not the fault of the people who write free software that most mainstream game developers seem to have no interest in supporting their platforms. On the otherhand, you say you tried Fedora Core. To make MP3 support, DVD playing, etc available to the package manager takes one command. Got to http://rpm.livna.org/ and see how long it takes you to find it... (and before you say Livna is obscure, it's easily found by Googling).
However, in saying that you want click-and-go install of applications without compiling, you're missing the point. This is only possible using a distribution's package management system, or distribution-specific click-and-go systems such as Klik. Why? The reason is that precompiled software is architecture- and platform-specific. A program compiled to run on x86 Slackware will never work on PPC Gentoo, and vice versa. In addition, consider that software seems to be written nowadays to run on either Windows, or anything except Windows. It will never be possible for developers to provide binaries for every possible combination of kernel+libraries, so they spend vast amounts of time writing e.g. autoconf scripts so that (hopefully) all you have to do is type "configure && make install" -- a process I don't consider to be too difficult to understand.[1][2]
I would take issue with your statement that working with Linux is like "[Figuring] out how the damn carborator (sic) works when I mainly want to drive." If you want to fiddle with the innards of Windows, you have to drill down into obscure, poorly documented configuration dialogs, and brave the mess that is the registry (why the hell should I have to enter binary data into an undocumented registry key in order to do something as simple as swap my Ctrl and Caps Lock keys?) If you want to fiddle with the innards of Linux, most things are configurable from reasonably-named plain-text configuration file somewhere on your system. Having said that, the last time I had to do fire up an editor and hack a config file was when I was trying to do something decidedly non-standard... several months ago.
I'd say that moving to Linux from Windows is like moving from driving a family car to driving an aircraft; although the controls are very different, once you've mastered them you get where you're going much more quickly and comfortably.
[1] Yes, I realize that bytecode-compiled languages and interpreted languages allow one binary to run on many systems. But they're mostly relatively slow.
[2] I agree that when "./configure" fails, it's often quite difficult to work out what's gone wrong. Many developers aren't good at writing meaningful error messages for the self-configuration stage, unfortunately.
That is not true, to a ridiculous extent. I use my Linux computer for some very obscure things: VHDL coding and simulation, CAD and low-level serial IO, some of which I do while logged in remotely. All of the tools I required for that were available to install using a simple graphical utility that took care of all the dependencies for me.
Lets add the fact that I had to reinstall my computer from scratch a month ago, and it took me:
You, sir, are a troll, who clearly hasn't tried a Linux distribution released since 2002.
Fedora Core 4 automounter works perfectly for me, out of the box. *shrug*
Guess why that is? You don't learn how to do that even at university. And no company will give you a job doing that if you don't already know how to do it, because they don't want to have to pay you while you learn. Not to mention the fact that it's impossible to do in your own time because the layout and simulation tools needed to create a working design are ridiculously expensive.
The problem is that creating good symbols is a bit of a black art, in choosing how to lay out the pins and whereabouts on the symbol to put the buses. I don't see anyone automating it in a sensible way any time soon, unfortunately.
PCB now comes with a whole bunch of parameterized footprints now, though, which are quite nice. Also, a number of footprint-generating scripts for various types of footprints are out there -- you just have to look for them. However, I haven't yet found a footprint that _wasn't_ already available in PCB...
...are they going to ban microwave ovens too?
There's a lot of professional grade software out there for professionals that runs as well or better on Linux than on any other operating system:
I could carry this list on for a long time, and that's just in my field. I keep on having to repeat myself: for high-end engineering research and development, people are upgrading to Linux all the time.
Cambridge University Engineering Department use Linux workstations for all undergraduate tuition -- they're based on a custom Knoppix distribution. More information at the CUED website.
Professional software for professionals: lots of it for Linux. Maybe you mean "professional" software for amateurs?
Yes, but my suspicion is that they consist of high-performance ADC at one end, a DAC at the other, an FFT, and an IFFT, with some digital logic tying them all together.
Most cutting edge signal processing is done in the digital domain: for instance, using FFTs it's possible to build a 100 kHz high-pass filter that has a stupidly high Q-factor.
Give them the finger, of course. Then post the e-mail to Slashdot. :P