The Berlin Wall stood between 1961 and 1989 (was totally gone by 1990). Link Here.
That's 28 years by my reckoning. Unix has lasted longer; Linux, whether it survives or not, hasn't had the chance last over 28 years yet.
Whether Linux as such is around in 5, 10, 28 years' time... maybe, maybe not. But the code behind it can be (may not be, if it's not good enough, but it can be) - unlike closed-source licensed software, Linux (and all GPL software) can evolve.
I'll skip some bits for those who have the time to totally deconstruct this post, but:
icons suggest direct functional correlations with real world objects and their intended computing task are very useful design elements (i.e. mailboxes for mail, telephones for modems, paintbrushes for graphic illustration, etc) :
All of the above examples you provide embody the problem with most current UIs (and, to an extent, gui applications). Taking them in turn:
Mailboxes for Mail: This is the best of the lost, but still flawed: There are huge differences between postal mail and email. Also, most GUIs use US-style mailboxes, whereas most software is distributed worldwide.
Telephones for Modems: Sure, you can use a phone line to connect to the internet; few people have used a telephone handset to connect in the past ten years!
Paintbrushes for Graphic Illustration: This is the worst of the lot. I don't know of an existing UI which allows one to use a paintbrush with a graphics program - if there was, I'd use that icon for it. As it is, I tend to use my mouse. Others may use a graphics tablet. There is nothing intuitive here
Everything you offer here is totally non-intuitive. Think: If you could transport someone from 1900 and ask them to guess what these icons would do, how accurate would their guesses be? 1% if you chose a real idiot; 0% if you chose anybody brighter.
All of the above require some knowledge of the limitaions of current computers (and to some extent, how PCs are better than the real-world equivalents).
If I know nothing about PCs, and want to let somebody get a message NOW, then Mail is not the concept I want to use.
Similarly, if I want "That internet thing", that's a different thing to me, than making a phone call.
And if I want to edit a photograph, I don't want to paint over it.
WinZip is distributed in this way because it's a promotional exercise, and it's worked - the.ZIP format is huge, and WinZip pretty much the only way for Windows users to access compressed files.
Naming no names, but having a monopoly is more important that the money you make from it, if you take a long-term view.
If I write buggy code and promote it as secure, then it is later found to be insecure, that is my fault, and mine alone. Same goes for MS, IBM, etc. But if somebody finds a flaw, that's still my problem, not theirs.
The script-kiddie issue is just one we must live with; A sysadmin can only get access to the same data as a script-kiddie can. If the sysadmin needs it, then with that right comes a responsibility to test and patch systems.
If non-disclosure is good for vendors and consumers, then *all* vendors would be pushing for non-disclosure - not just Microsoft. As it is, the *nix vendors are happy with the status quo, because they are reasonably happy with the quality of their software.
However, proof-of-concept code does not necessarily prove/disprove the presence of a threat. If the exploit addresses a particular subset of an overall fault, then just because the exploit fails does not mean that the fault has been fixed - see compiler optimisation "cheats" for example.
When the source is open, then the exploit can be more easily shown to be complete/incomplete - otherwise, the exploit either works/doesn't work - but that could imply a failing in the exploit code, or a partial workaround in the patch, which simply avoids the exploit.
I disagree with the author that there is no incentive for vendors to provide secure code - Secure Solaris being one example - but the customer must pay megabucks for such a promise, and compromise the customisability they expect.
In his essay, Culp compares the practice of publishing vulnerabilities to shouting "Fire" in a crowded
movie theater. What he forgets is that there actually is a fire, the vulnerabilities exist regardless.
Blaming the person who disclosed the vulnerability is like imprisoning the person who first saw the
flames.
Nuff Said.
And no journalling filesystem can ensure data integrity if the power dies... the memory can fill with all kinds of crap, which may or may not get flushed to disk. This is true of all PCs, and basically any Von-Neumann system (ie, any computer).
Let me give an example. Recently I was in a "computer centre" of a certain organisation (I work for Sun, so we tend to expect datacentres of a certain 'class'). Far from the normal - racks and racks of Compaq, Sun, EMC, and some IBM, there were small tables all over the place, mini-tower PCs with crappy monitors, mice and keyboards attached to them, each with their own UPS (but no serial cable, even!), each one running another instance of Windows NT.
This is what happens when you do not plan properly:
We need a mail server: We'll put a new PC in to do that
We need a web server: We'll put a new PC in to do that
etc.etc.etc.
So this place was stacked to the gills with PCs, each needing a mouse and monitor, running NT. Quite a few had post-it notes on the monitors, saying things like "leave this PC logged in as app-admin", etc.
Decent planning would have had a few large machines, well powered, cooled, and administrated.
And probably running some kind of *nix. Frankly, I don't care if they're on Linux, Solaris, even some mythical stable OS from Microsoft. Bad planning really p*sses me off.
Does the new VM mean that StarOffice loads any quicker? It thrashes for about 10 seconds under MS Windoze before being ready, 12 under Solaris_x86, and 30 under Linux 2.2.18 or 2.4.7
Yeah, there are some high-falutin' OS concepts flying around here, but what really matters to me is, does it actually affect what I'm doing? Not a huge database, but if it can stop StarOffice thrashing so much (and it must be Linux doing it, since Solaris_x86 is so much quicker), then maybe I'll upgrade my kernel.
(All timings taken on identical 433MHz Celerons, 128MB, single 6MB IDE disk)
Gnome 1.4 is available on the extras CD which comes with Solaris 8, along with a whole bunch of G-Goodies (gcc, cdrecord, ghostview, it's a huge list, the whole lot it listed at www.sun.com.
Don't bitch about the RIAA when they push for freedom-curtailing laws - THEY DON'T CARE - their job is
to protect the interests of those who pay their salaries.
Now I'd call that one hit for AOL, and one each for the celebrity sites, but AOL acutally racks up 5 hits here, or 5/7 = 71percent of the traffic. So users are using www.aol.com as their homepage, cause they don't know how to change it, and search.aol.com as their search engine, cause they don't know there's anything else.
I'd like to see this for Sparc architectures...I have one of the Sparc clones with an ATI Rage II+ in it. 256 colors is possible, but just you try to set different video modes...anyone have any ideas for me? The board is PCI-based, and I'm not paying 700 bucks for a video card that a TNT1 would stomp all over...>
Check out infodoc 18511 from sunsolve
By default, Ultra 5's with the PGX24 graphics card start up in 8-bit color
depth mode. To change the color depth to 24 bits, do the following:
1. Install the patch 103792-14 or later if you are running Solaris 2.5.1 11/97
Install the patch 105362-13 or later if you are running Solaris 2.6 5/98
2. Reboot the machine after installing the patch above.
You cannot go higher than 1152x900 in 24 bit mode on the PGX24 framebuffer.
#/usr/dt/bin/dtlogin -deamon; exit
-- or reboot the system.
The above steps are needed so that Openwindows/CDE Xsun server does not write over the m64config settings. You cannot effectively use the m64config command while Xsun is running on the system.
4. If you are in CDE or Openwindows, please exit now and login again
so the depth changes take place.
5. Now run/usr/openwin/bin/xwininfo and select the background. You should
see the depth set to 24 if the Xsun was started in 24 bit mode in
/etc/dt/config/Xservers (defdepth 24). If it was set to 8 bit mode (defdepth 8),
you will see that this file's settings overwrite the m64config settings.
In other words, in CDE/Openwindows you will see:
# m64config -prconf |grep Current
Current resolution setting: 1152x900x76
Current depth: 8
Also, see SRDB 18439 (why doesn't my PGX+ card keep it's 24-bit color depth
setting?).
How much of 'big-iron management' can you learn on an Intel box, without the right hardware?
I use Linux on my laptop, and it's highly inconvenient when I want to develop shell scripts for Solaris boxes. But Linux supports my hardware better, so it's a compromise. But I'd love to be able to run Solaris on the hardware I've got - that would be the ideal solution.
Also, things like the SunSolve interactive CD is for Solaris, so I can't run that under Linux.
Surely there must be _somebody_ who is using Solaris x86 for real work?
Yes - we're a Sun partner, and use Solaris_x86 for our webserver and an application server. We've got SPARC (Solaris) workstations, and laptops which are either Linux, Windows and dual-boot Win/Linux.
And has been said, we use x86 because it's cheap, and we use Solaris because it's good.
What exactly is your objection to this term?
It's too thin to be a client? So what would you call a thin client?
do we really want to move back to the old time-sharing model, where you can't do anything without the approval of computer center
In a working environment, yes. What objection can you possibly have to this approach?
There is no need for a seceratary to have a machine which can do any more than create documents. Sure, developers need a decent machine to work on.
When you're older, you'll learn to differentiate between the two, and realise that nobody's taking your toys away from you without good reason.
In the interview, Michels shows a big misunderstanding in what he has bought.
He seems to believe that he has Linux, and has now bought Unix, and wants to bring these together. This seems - to me - to be the strongest case yet for the popularisation of the term "GNU/Linux", since all he seems to be saying is combining the GNU stuff which comes with Linux distros with "his" Unix kernel.
All he is proposing is GNU/SysV as opposed to GNU/Linux or even GNU/Hurd (if that ever happens).
So in the good ol' days, AT&T owned Unix. Then various companies licensed it from them, and that's how the various flavours formed.
IBM licensed SysV from AT&T, and created AIX; HP created HP-UX; Sun created SunOS.
But what *was* the licensing deal? I'm sure we're not dealing with a one-off payment to AT&T.
So do these companies (IBM, HP, Sun, etc) now have to pay Caldera for the use of SysV-based Unices? I've heard nothing along these lines, and surely the big guys would have something to say if it was the case.
So what does it matter that Caldera own SysV? It wasn't written for Intel anyway, and what counts in today's world (for the x86 anyway) is hardware support. The big guys build their own hardware, and their flavour of Unix supports it.
So who really cares if Caldera, or Fred Smith down the road owns the rights to SysV?
This is complete bullshit. MS will never charge for IE. There is no point in doing so - pretty much nobody else charges for their web browser.
That was the point in giving IE away for free. Kill the competition, IE owns, what, 80% of the browser market?
Job Done
All that remains is to kill off these pesky outsiders - RealMedia for allowing users to get AV without MS software, Sun for allowing users to run Java without MS software, Apple for.... you get the picutre.
Slashdot Mirror
Thank God for that ;-)
No - spam this account with idiotic pro-MS rants, so that all pro-MS rants will be ignored
;-)
;-)
That's 28 years by my reckoning. Unix has lasted longer; Linux, whether it survives or not, hasn't had the chance last over 28 years yet. ... maybe, maybe not. But the code behind it can be (may not be, if it's not good enough, but it can be) - unlike closed-source licensed software, Linux (and all GPL software) can evolve.
Whether Linux as such is around in 5, 10, 28 years' time
I'll skip some bits for those who have the time to totally deconstruct this post, but:
icons suggest direct functional correlations with real world objects and their intended computing task are very useful design elements (i.e. mailboxes for mail, telephones for modems, paintbrushes for graphic illustration, etc) :
All of the above examples you provide embody the problem with most current UIs (and, to an extent, gui applications). Taking them in turn:
Everything you offer here is totally non-intuitive. Think: If you could transport someone from 1900 and ask them to guess what these icons would do, how accurate would their guesses be? 1% if you chose a real idiot; 0% if you chose anybody brighter.
All of the above require some knowledge of the limitaions of current computers (and to some extent, how PCs are better than the real-world equivalents).
If I know nothing about PCs, and want to let somebody get a message NOW, then Mail is not the concept I want to use.
Similarly, if I want "That internet thing", that's a different thing to me, than making a phone call.
And if I want to edit a photograph, I don't want to paint over it.
As Etyenne says, it's nothing original, but
IT'S TRUE
(If that matters on
That's the whole speech/beer thing.
.ZIP format is huge, and WinZip pretty much the only way for Windows users to access compressed files.
WinZip is distributed in this way because it's a promotional exercise, and it's worked - the
Naming no names, but having a monopoly is more important that the money you make from it, if you take a long-term view.
If you want to ship anything, check the terms and conditions first.
... for exactly this reason.
Most carriers are not insured to carry computing equipment anyway; those who are, will charge extra for it
You get what you pay for, and you didn't pay.
It may not be nice, but it's life.
Your monotheistic religions are fucking stupid.
All I want for Christmas is Osama bin Laden's head. Is this irony intentional?
The script-kiddie issue is just one we must live with; A sysadmin can only get access to the same data as a script-kiddie can. If the sysadmin needs it, then with that right comes a responsibility to test and patch systems.
If non-disclosure is good for vendors and consumers, then *all* vendors would be pushing for non-disclosure - not just Microsoft. As it is, the *nix vendors are happy with the status quo, because they are reasonably happy with the quality of their software.
However, proof-of-concept code does not necessarily prove/disprove the presence of a threat. If the exploit addresses a particular subset of an overall fault, then just because the exploit fails does not mean that the fault has been fixed - see compiler optimisation "cheats" for example.
When the source is open, then the exploit can be more easily shown to be complete/incomplete - otherwise, the exploit either works/doesn't work - but that could imply a failing in the exploit code, or a partial workaround in the patch, which simply avoids the exploit.
I disagree with the author that there is no incentive for vendors to provide secure code - Secure Solaris being one example - but the customer must pay megabucks for such a promise, and compromise the customisability they expect.
In his essay, Culp compares the practice of publishing vulnerabilities to shouting "Fire" in a crowded movie theater. What he forgets is that there actually is a fire, the vulnerabilities exist regardless. Blaming the person who disclosed the vulnerability is like imprisoning the person who first saw the flames.
Nuff Said.
RedHat can run an SMP Kernel out-of-the box.
And no journalling filesystem can ensure data integrity if the power dies... the memory can fill with all kinds of crap, which may or may not get flushed to disk. This is true of all PCs, and basically any Von-Neumann system (ie, any computer).
How's the weather on your planet?
Let me give an example. Recently I was in a "computer centre" of a certain organisation (I work for Sun, so we tend to expect datacentres of a certain 'class'). Far from the normal - racks and racks of Compaq, Sun, EMC, and some IBM, there were small tables all over the place, mini-tower PCs with crappy monitors, mice and keyboards attached to them, each with their own UPS (but no serial cable, even!), each one running another instance of Windows NT.
This is what happens when you do not plan properly:
We need a mail server: We'll put a new PC in to do that
We need a web server: We'll put a new PC in to do that
etc.etc.etc.
So this place was stacked to the gills with PCs, each needing a mouse and monitor, running NT. Quite a few had post-it notes on the monitors, saying things like "leave this PC logged in as app-admin", etc.
Decent planning would have had a few large machines, well powered, cooled, and administrated.
And probably running some kind of *nix. Frankly, I don't care if they're on Linux, Solaris, even some mythical stable OS from Microsoft. Bad planning really p*sses me off.
Does the new VM mean that StarOffice loads any quicker? It thrashes for about 10 seconds under MS Windoze before being ready, 12 under Solaris_x86, and 30 under Linux 2.2.18 or 2.4.7
Yeah, there are some high-falutin' OS concepts flying around here, but what really matters to me is, does it actually affect what I'm doing? Not a huge database, but if it can stop StarOffice thrashing so much (and it must be Linux doing it, since Solaris_x86 is so much quicker), then maybe I'll upgrade my kernel.
(All timings taken on identical 433MHz Celerons, 128MB, single 6MB IDE disk)
Gnome 1.4 is available on the extras CD which comes with Solaris 8, along with a whole bunch of G-Goodies (gcc, cdrecord, ghostview, it's a huge list, the whole lot it listed at www.sun.com.
Don't bitch about the RIAA when they push for freedom-curtailing laws - THEY DON'T CARE - their job is
to protect the interests of those who pay their salaries.
But WE pay their salaries!!!
Consider a typical session where an AOL user is looking for information on Natalie Portman.
1. www.aol.com (homepage)
2. search.aol.com
3. search.aol.com?Natalie Portman
4. www.adoredcelebrities.com
5. search.aol.com
6. search.aol.com?Natalie portman
7. www.anycelebrity.com
Now I'd call that one hit for AOL, and one each for the celebrity sites, but AOL acutally racks up 5 hits here, or 5/7 = 71percent of the traffic. So users are using www.aol.com as their homepage, cause they don't know how to change it, and search.aol.com as their search engine, cause they don't know there's anything else.
I'd like to see this for Sparc architectures...I have one of the Sparc clones with an ATI Rage II+ in it. 256 colors is possible, but just you try to set different video modes...anyone have any ideas for me? The board is PCI-based, and I'm not paying 700 bucks for a video card that a TNT1 would stomp all over...>
/usr/bin/ps -ef|/usr/bin/grep -v grep |/usr/bin/grep -v ksh|/usr/bin/grep dtlogin| /usr/bin/awk '{print $2}'|xargs kill -9
/usr/dt/bin/dtlogin -deamon; exit
/usr/openwin/bin/xwininfo and select the background. You should
/etc/dt/config/Xservers (defdepth 24). If it was set to 8 bit mode (defdepth 8),
Check out infodoc 18511 from sunsolve
By default, Ultra 5's with the PGX24 graphics card start up in 8-bit color
depth mode. To change the color depth to 24 bits, do the following:
1. Install the patch 103792-14 or later if you are running Solaris 2.5.1 11/97
Install the patch 105362-13 or later if you are running Solaris 2.6 5/98
2. Reboot the machine after installing the patch above.
3. Login as root and run the following command:
Kill dtlogin
#
#m64config -depth 24
Check your settings:
# m64config -prconf |grep Current
Current resolution setting: 1152x900x76
Current depth: 24
You cannot go higher than 1152x900 in 24 bit mode on the PGX24 framebuffer.
#
-- or reboot the system.
The above steps are needed so that Openwindows/CDE Xsun server does not write over the m64config settings. You cannot effectively use the m64config command while Xsun is running on the system.
4. If you are in CDE or Openwindows, please exit now and login again
so the depth changes take place.
5. Now run
see the depth set to 24 if the Xsun was started in 24 bit mode in
you will see that this file's settings overwrite the m64config settings.
In other words, in CDE/Openwindows you will see:
# m64config -prconf |grep Current
Current resolution setting: 1152x900x76
Current depth: 8
Also, see SRDB 18439 (why doesn't my PGX+ card keep it's 24-bit color depth
setting?).
How much of 'big-iron management' can you learn on an Intel box, without the right hardware?
I use Linux on my laptop, and it's highly inconvenient when I want to develop shell scripts for Solaris boxes. But Linux supports my hardware better, so it's a compromise. But I'd love to be able to run Solaris on the hardware I've got - that would be the ideal solution.
Also, things like the SunSolve interactive CD is for Solaris, so I can't run that under Linux.
Surely there must be _somebody_ who is using Solaris x86 for real work?
Yes - we're a Sun partner, and use Solaris_x86 for our webserver and an application server. We've got SPARC (Solaris) workstations, and laptops which are either Linux, Windows and dual-boot Win/Linux.
And has been said, we use x86 because it's cheap, and we use Solaris because it's good.
You have always been able to run XFree86 on Solaris_x86. This has no impact whatsoever on Sol_x86's usability.
It's too thin to be a client? So what would you call a thin client?
do we really want to move back to the old time-sharing model, where you can't do anything without the approval of computer center
In a working environment, yes. What objection can you possibly have to this approach?
There is no need for a seceratary to have a machine which can do any more than create documents. Sure, developers need a decent machine to work on.
When you're older, you'll learn to differentiate between the two, and realise that nobody's taking your toys away from you without good reason.
What the guy fails to understand, is that Linux and Unix cannot co-exist - a system can only run one kernel, be that Unix or Linux.
What he seems to be pushing is a Unix box which can imitate a Linux box if necessary.
Now why would I put my budget towards his solution, and not a GNU/Linux one?
In the interview, Michels shows a big misunderstanding in what he has bought.
He seems to believe that he has Linux, and has now bought Unix, and wants to bring these together. This seems - to me - to be the strongest case yet for the popularisation of the term "GNU/Linux", since all he seems to be saying is combining the GNU stuff which comes with Linux distros with "his" Unix kernel.
All he is proposing is GNU/SysV as opposed to GNU/Linux or even GNU/Hurd (if that ever happens).
So in the good ol' days, AT&T owned Unix. Then various companies licensed it from them, and that's how the various flavours formed.
IBM licensed SysV from AT&T, and created AIX; HP created HP-UX; Sun created SunOS.
But what *was* the licensing deal? I'm sure we're not dealing with a one-off payment to AT&T.
So do these companies (IBM, HP, Sun, etc) now have to pay Caldera for the use of SysV-based Unices? I've heard nothing along these lines, and surely the big guys would have something to say if it was the case.
So what does it matter that Caldera own SysV? It wasn't written for Intel anyway, and what counts in today's world (for the x86 anyway) is hardware support. The big guys build their own hardware, and their flavour of Unix supports it.
So who really cares if Caldera, or Fred Smith down the road owns the rights to SysV?
That was the point in giving IE away for free. Kill the competition, IE owns, what, 80% of the browser market?
Job Done
All that remains is to kill off these pesky outsiders - RealMedia for allowing users to get AV without MS software, Sun for allowing users to run Java without MS software, Apple for .... you get the picutre.