One of the issues I have been trying to understand about the GPL is exactly when you are or are not allowed to link to non-GPL-compliant libraries. IANAL, if you want a legal advice buy it from a licensed attourney, etc.
See the penultimate paragraph of clause 3:
However, as a special exception,
the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
This means that distributing a Solaris binary of a GPLed program that links against Solaris' proprietary C library is fine. Distributing a binary of a GPLed program that links against liboracle.so (or liboracle.a, for that matter) isn't, though - unless all the copyright owners of the GPLed package explicitly agree and license it to you under a seperate, modified license that permits such linking.
If you read the general overviews, it would seem that such linking is always explicitly prohibited. But the license doesn't mention anything of the sort. I have not been able to find the word "link" used in this context at all in the license. Here are a few common options I have thought of that might contradict the "in a nutshell" overviews of the license:
1) I download GPL'd code and compile it for my own use (no redistribution) and I use a proprietary ANSI C library. This is probably a common issue on Windows where you may be linking to the Windows ANSI C libraries for a native executable.
Do what you like. If there's no distribution, the GPL doesn't have anything to say on the matter.
2) I download GPL'd code and it has a wrapper which interfaces via linking with a number of proprietary packages (think NDIS-wrapper), or perhaps it allows another GPL'd project to link to this using standard interfaces.
If you were creating such a package, to be technically correct, you probably need to specifically mention the proprietary objects you allow your code to be linked against. In the case of NDIS-wrapper, I imagine the name is good enough. Now, whilst distributing a binary of the Linux kernel with NDIS-wrapper compiled in is fine, there's nothing to stop Linus and co. complaining if someone distributes a bunch of NDIS drivers with it. Just like with 1) above, though, there's nothing to stop you, as the end user from adding whatever NDIS drivers you wish to your machines (modulo any terms in the licenses for those drivers, of course!)
Red Hat Enterprise Linux on important servers. Guaranteed updates until 2010 makes me happy.
CentOS or Whitebox on servers that are less important, or more numerous than the notes in the corporate kitty. Both are Free rebuilds of RHEL that should have similar update availability, compatibility and QA.
Fedora on expert users' workstations or servers that need the latest and greatest TODAY, and damn the consequences.
I bought the ATI Radeon 9800 and I was terribly disappointed. The fan fried not once, twice, but 3 times. I had paid almost $100 in RMA returns and shipping.
As another poster pointed out; whose video card did you buy? I would guess that it was some third party's, not ATI-on-ATI (they don't seem to build many these days; shame, because they used to be very well built). Personally, I've found that the passive components on my cheap Radeon 7500 give better picture quality than the cheap Geforce 440MX I bought at the same time for about the same price.
I no longer can leave my PC running around the clock because I know the card would fry if I leave it up. I already have gigantic fans running with open case. No overclocking at all.
Generally, you get better cooling if you have your case closed than open. Ideally, you want cool air coming in at the bottom of the case, and warm air being extracted from the top. Chances are with your setup that those big fans are just swirling the air around a bit, rather than creating forced convection.
Sorry ATI, but I am going back to Nvidia in my next purchase. ATI drivers are also terribly lousy. If you need a new Catalyst driver every month, you got problems. Half the games were always filled with overheated white dots. I treat my hardware with RTFM care. And I deal with another ATI product again.
nVidia's drivers suck too. They've just broken again with kernel 2.6.10. I've got some patches that I'm meaning to try which claim to fix it, but it's really not a priority.
Indeed - I used to play UT on my Celeron 500/Rage 128 pro, and I play RTCW on my P4 2.4/Radeon 7500LE. Both are quite playable (though RTCW did crash once, at the "burning guy rising out of the ground" stage).
I'm sure hardcore gamers would have some unpleasant things to say, but it's my money, not theirs!:-)
I'd love to buy a modern video card with OSS drivers
How modern do you need? The Radeon 9200/9250 is supported out-of-the-box for accelerated 3D by a recent kernel + xorg. If you don't care about 3D, then the sky's the limit, pretty much.
the vast majority of people who buy computers really honestly couldn't give a crap about [...] "small form factor".
Not to diminish your other points, but actually, they do; a few of my married friends have been banned from bringing any more beige boxes into the marital home.;-]
I've put a mains power meter inline with my main desktop PC (350W PSU, P4 2.4, Radeon 7500, 512M RAM, 2x80G 7200 RPM discs, 1x DVD-ROM drive, 1x CD-RW drive, TV card) and whilst the meter shows it drawing about 100-110W at boot, at idle it only draws about 70-80W.
What really worries me (and perhaps this is unwarranted worrying) is the fact that every desktop Linux install I've ever seen is really patchworked... dozens of apps written by different people who have never even taked or discussed guidelines and several places to make system configuration settings.
This is pretty much inherent in the approach that Linux distros take. The BSDs seem to do things in a more co-ordinated way, and Apple seems to do things better again. That said, if you take the trouble to understand the changes Red Hat/Fedora (especially) make, you'll find they're actually working hard to create a sane system. Some things aren't in place, yet, though.
I think Microsoft's concern was, will RedHat be the ones to develop and push out a code fix for a random open source app that was included with their distro?
If the community comes up with the fix first (perhaps by releasing a new upstream version), Red Hat will include it, or backport the fix from the new upstream version. If Red Hat are first to fix it, their fix will go in - and probably also into the upstream release during the testing phase of the current version of the distro.
I think they have the framework to push out the fix, but probably not to fix the app itself.
Eh? What's that supposed to mean? If it's fixed, it's fixed.
Care to elaborate? Just what part of the software stack is missing?
The Man from Microsoft said 'no single development environment' (quite right, we have many, pick the one that suits your organisation best and standardise on it) and 'no single sign-on system' (er, what about Kerberos? RH/FC can even integrate with Microsoft Active Directory if you wish!).
I think you'll find that's exactly the point the OP (tdemark) was making.
Red Hat takes reponsibility for their distro in the same way Evian takes reponsibility for the safety of the water they sell. But neither take responsibility for all instances of the raw materials they package and sell.
You said business desktop, right? Really, Linux on the desktop is fine and dandy for us nerds, but it's not OK for business-at-large. Your average office worker knows Word, Excel, etc., not OOo or gnumeric.
I dispute this. most of the 'average office workers' neither know nor use enough of the esoteric functionality of Word, Excel etc. to make any difference what application they use, as long as it has the basic functionality and buttons/menus in roughly the same place.
Now, Word/Excel power-users are a different kettle of fish. Which is why the better-planned migrations seem to be keeping such users on MS Office (for now, at least).
"My" thinkpad uses the touchpad as the "mouse wheel"... vertical AND horizontal. Lotsa fun! I rarely use the middle button that's provided. Not sure about with Linux, but (Windoze)...
USians is short for the US and asian countries that lock down their currency to ours.
it is derogatory.
As a UKian, I've always taken it as a shorthand to mean '[citizen] of the United States of America'. 'American' is too vague, as that includes Canada and the nations of South America.
As ridiculous region coding is for DVDs, there I can see a minimal reason (the publishers not wanting a DVD to make it into a market where the movie hasn't even been in the cinemas yet... But as cinema release dates for the big global productions inch ever closer to each other all over the globe, this reason is going away fast - leaving the only "good" thing of the region codings that they can charge more in Europe.
Not only that, but I decided the ostensible justification for region coding (i.e. above) was bogus when I saw a region-coded DVD of Ghostbusters - some 17 years after it came out in the cinema!
Just as some software is built with the attention to safety that Volvo and Saab are famed for (e.g. OpenBSD), other software is built much like the Ford Pinto was. Users keep picking the Pinto-equivalents because it's cheaper (specifically: cheaper from a convenience and familiarity point-of-view, rather than in monetary terms). Manufacturers of Pinto-software keep producing it because customers keep buying it and there's no laws that prevent them from selling it.
I see your point but there is an exception: all of those things can be turned off.
$ su - Password: #/sbin/service iptables stop # echo 'password' | passwd --stdin root #/sbin/chkconfig telnet on #/sbin/service xinetd restart
Any computer system can be intentionally weakened. You may need more expertise than I've displayed above (cf. showing your racing license), but it's always possible.
Maybe BMW have spoilt you, but with most mass-market cars (for approximately the same market as the SimPC is intended, rather than driving enthusiasts or hackers), it's extremely rare that any of those features can be disabled, even though ABS is actually sub-optimal for safety in some conditions (i.e. snow).
And instead of fixing their own damn deffective product, the computer industry keeps blaming the user. "Noo, our product is perfect. It's those idiots who broke it. Let's give them a crippled locked-down PC they can't break." It's an idiocy of the calibre of "let's give them a car which only goes in a straight line, so they don't break it by taking tight curves."
Cars that only go in straight lines are bit too limiting for the average traveller, but how about rev limiters? speed limiters? power-assisted steering? traction control? ABS braking? All instances of cars being 'crippled and locked-down' compared with the 1900 models on one hand, but important safety improvements on the other.
A friend (who writes games, as it happens) reported to me that after a day go-karting, he found it quite hard to drive home in a safe and legal manner.
Similarly, I found that when I had a crack at racing Pilot buggies, it felt like I was just playing a driving sim. Well, until I span out, forgot to let go of the steering wheel and dislocated my shoulder, anyway. That felt pretty real.
Slashdot Mirror
One of the issues I have been trying to understand about the GPL is exactly when you are or are not allowed to link to non-GPL-compliant libraries. IANAL, if you want a legal advice buy it from a licensed attourney, etc.
See the penultimate paragraph of clause 3:
This means that distributing a Solaris binary of a GPLed program that links against Solaris' proprietary C library is fine. Distributing a binary of a GPLed program that links against liboracle.so (or liboracle.a, for that matter) isn't, though - unless all the copyright owners of the GPLed package explicitly agree and license it to you under a seperate, modified license that permits such linking.
If you read the general overviews, it would seem that such linking is always explicitly prohibited. But the license doesn't mention anything of the sort. I have not been able to find the word "link" used in this context at all in the license. Here are a few common options I have thought of that might contradict the "in a nutshell" overviews of the license: 1) I download GPL'd code and compile it for my own use (no redistribution) and I use a proprietary ANSI C library. This is probably a common issue on Windows where you may be linking to the Windows ANSI C libraries for a native executable.
Do what you like. If there's no distribution, the GPL doesn't have anything to say on the matter.
2) I download GPL'd code and it has a wrapper which interfaces via linking with a number of proprietary packages (think NDIS-wrapper), or perhaps it allows another GPL'd project to link to this using standard interfaces.
If you were creating such a package, to be technically correct, you probably need to specifically mention the proprietary objects you allow your code to be linked against. In the case of NDIS-wrapper, I imagine the name is good enough. Now, whilst distributing a binary of the Linux kernel with NDIS-wrapper compiled in is fine, there's nothing to stop Linus and co. complaining if someone distributes a bunch of NDIS drivers with it. Just like with 1) above, though, there's nothing to stop you, as the end user from adding whatever NDIS drivers you wish to your machines (modulo any terms in the licenses for those drivers, of course!)
Red Hat Enterprise Linux on important servers. Guaranteed updates until 2010 makes me happy.
CentOS or Whitebox on servers that are less important, or more numerous than the notes in the corporate kitty. Both are Free rebuilds of RHEL that should have similar update availability, compatibility and QA.
Fedora on expert users' workstations or servers that need the latest and greatest TODAY, and damn the consequences.
As another poster pointed out; whose video card did you buy? I would guess that it was some third party's, not ATI-on-ATI (they don't seem to build many these days; shame, because they used to be very well built). Personally, I've found that the passive components on my cheap Radeon 7500 give better picture quality than the cheap Geforce 440MX I bought at the same time for about the same price.
I no longer can leave my PC running around the clock because I know the card would fry if I leave it up. I already have gigantic fans running with open case. No overclocking at all.
Generally, you get better cooling if you have your case closed than open. Ideally, you want cool air coming in at the bottom of the case, and warm air being extracted from the top. Chances are with your setup that those big fans are just swirling the air around a bit, rather than creating forced convection.
Sorry ATI, but I am going back to Nvidia in my next purchase. ATI drivers are also terribly lousy. If you need a new Catalyst driver every month, you got problems. Half the games were always filled with overheated white dots. I treat my hardware with RTFM care. And I deal with another ATI product again.
nVidia's drivers suck too. They've just broken again with kernel 2.6.10. I've got some patches that I'm meaning to try which claim to fix it, but it's really not a priority.
I'm sure hardcore gamers would have some unpleasant things to say, but it's my money, not theirs! :-)
How modern do you need? The Radeon 9200/9250 is supported out-of-the-box for accelerated 3D by a recent kernel + xorg. If you don't care about 3D, then the sky's the limit, pretty much.
Not to diminish your other points, but actually, they do; a few of my married friends have been banned from bringing any more beige boxes into the marital home. ;-]
I've put a mains power meter inline with my main desktop PC (350W PSU, P4 2.4, Radeon 7500, 512M RAM, 2x80G 7200 RPM discs, 1x DVD-ROM drive, 1x CD-RW drive, TV card) and whilst the meter shows it drawing about 100-110W at boot, at idle it only draws about 70-80W.
This is pretty much inherent in the approach that Linux distros take. The BSDs seem to do things in a more co-ordinated way, and Apple seems to do things better again. That said, if you take the trouble to understand the changes Red Hat/Fedora (especially) make, you'll find they're actually working hard to create a sane system. Some things aren't in place, yet, though.
I think Microsoft's concern was, will RedHat be the ones to develop and push out a code fix for a random open source app that was included with their distro?
If the community comes up with the fix first (perhaps by releasing a new upstream version), Red Hat will include it, or backport the fix from the new upstream version. If Red Hat are first to fix it, their fix will go in - and probably also into the upstream release during the testing phase of the current version of the distro.
I think they have the framework to push out the fix, but probably not to fix the app itself.
Eh? What's that supposed to mean? If it's fixed, it's fixed.
The Man from Microsoft said 'no single development environment' (quite right, we have many, pick the one that suits your organisation best and standardise on it) and 'no single sign-on system' (er, what about Kerberos? RH/FC can even integrate with Microsoft Active Directory if you wish!).
Red Hat takes reponsibility for their distro in the same way Evian takes reponsibility for the safety of the water they sell. But neither take responsibility for all instances of the raw materials they package and sell.
How about 4 spindles of [S]ATA, RAID1[0]'ed vs. a single SCSI spindle of the same size as the array for the same price?
I've not benchmarked it, but I would expect the [S]ATA to be faster AND more reliable, given WD or Seagate discs.
I dispute this. most of the 'average office workers' neither know nor use enough of the esoteric functionality of Word, Excel etc. to make any difference what application they use, as long as it has the basic functionality and buttons/menus in roughly the same place.
Now, Word/Excel power-users are a different kettle of fish. Which is why the better-planned migrations seem to be keeping such users on MS Office (for now, at least).
Yup, it's available in Linux too. FC3 even includes it out of the box, though if you have an ALPS touchpad (common in many laptops, including my Toshiba Satellite 3000), you need to rebuild the kernel with a patch applied from the synaptics package
it is derogatory.
As a UKian, I've always taken it as a shorthand to mean '[citizen] of the United States of America'. 'American' is too vague, as that includes Canada and the nations of South America.
Not only that, but I decided the ostensible justification for region coding (i.e. above) was bogus when I saw a region-coded DVD of Ghostbusters - some 17 years after it came out in the cinema!
Alternatively, might it be intended that you use it together with the sulphur to make primitive black powder-based rescue flares? (blue!) --
I'd put good money on it being China and/or India.
Sadly, libraries aren't similarly short of security problems. This is what happens when a library that is commonly statically linked is found to have a security vulnerability.
This hasn't always been the case.
Just as some software is built with the attention to safety that Volvo and Saab are famed for (e.g. OpenBSD), other software is built much like the Ford Pinto was. Users keep picking the Pinto-equivalents because it's cheaper (specifically: cheaper from a convenience and familiarity point-of-view, rather than in monetary terms). Manufacturers of Pinto-software keep producing it because customers keep buying it and there's no laws that prevent them from selling it.
Maybe BMW have spoilt you, but with most mass-market cars (for approximately the same market as the SimPC is intended, rather than driving enthusiasts or hackers), it's extremely rare that any of those features can be disabled, even though ABS is actually sub-optimal for safety in some conditions (i.e. snow).
Cars that only go in straight lines are bit too limiting for the average traveller, but how about rev limiters? speed limiters? power-assisted steering? traction control? ABS braking? All instances of cars being 'crippled and locked-down' compared with the 1900 models on one hand, but important safety improvements on the other.
I've even heard the prelude to a call on live TV broadcasts (e.g. interviews) before now.
Similarly, I found that when I had a crack at racing Pilot buggies, it felt like I was just playing a driving sim. Well, until I span out, forgot to let go of the steering wheel and dislocated my shoulder, anyway. That felt pretty real.
Well, the one I linked to explicitly claims Mac OS support, so if it doesn't work, at least you won't have any problems getting your money back.
Just use of of these - two PS/2 devices into one USB port. You can probably find them at a computer fair for much less than Maplin sell them for, too.