On Thu, 2004-09-30 at 16:25, Enrico Scholz wrote:
documentation and/or diagnostic. And btw... the nvidia driver are not supported by FC ;)
Not sure what you mean by "supported" but NVIDIA-Linux-x86-1.0-5336-pkg1 works perfectly fine with FC2 for me...
Malita, Florin wrote:
On Thu, 2004-09-30 at 16:25, Enrico Scholz wrote:
documentation and/or diagnostic. And btw... the nvidia driver are not supported by FC ;)
Not sure what you mean by "supported" but NVIDIA-Linux-x86-1.0-5336-pkg1 works perfectly fine with FC2 for me...
Also that "not supported" thing is crap. Binary modules are a fact of life and some care ("some" meaning "break it if you have to, but avoid it if you can") must be taken so that users can use them.
The fact is that nvidia cards are _very_ popular and users want to have them working properly on linux. In fact they don't have much options, ATi drivers suck beyond reason and other cards are just plain insufficient preformance-wise. So, if the nvidia driver breaks and isn't easy to fix, that will just drive users away. Take me for an example, I have always updated to a newer Red Hat version (since Red Hat 5.0) whithin 1.5 weeks of release, but it took me around 2 months to step up to Fedora 2. Why? Because the nvidia drivers didn't work with the fedora 2 kernel and I have better things to do with my time that patching and rebuilding kernels. When nvidia released new drivers I upgraded within 1 week. If I were a new user, I wouldn't even have bothered.
Carlos Rodrigues
It wasn't a so big problem THEN - all i did was download a new kernel RPM from the ndiswrapper side with 8k stacks disabled.
*pof* nVidia!
But this udev stuff would be worse to work around... => my main machine will stay on fc2 untill nvidia drivers work with fc3. BTW pissing off nvidia isn't such a good idea either. Not that i oppose udev, but you get the idea. Please not break the drivers for fun. Not that i think you do.
Kyrre
fre, 01.10.2004 kl. 00.20 skrev Carlos Rodrigues:
Malita, Florin wrote:
On Thu, 2004-09-30 at 16:25, Enrico Scholz wrote:
documentation and/or diagnostic. And btw... the nvidia driver are not supported by FC ;)
Not sure what you mean by "supported" but NVIDIA-Linux-x86-1.0-5336-pkg1 works perfectly fine with FC2 for me...
Also that "not supported" thing is crap. Binary modules are a fact of life and some care ("some" meaning "break it if you have to, but avoid it if you can") must be taken so that users can use them.
The fact is that nvidia cards are _very_ popular and users want to have them working properly on linux. In fact they don't have much options, ATi drivers suck beyond reason and other cards are just plain insufficient preformance-wise. So, if the nvidia driver breaks and isn't easy to fix, that will just drive users away. Take me for an example, I have always updated to a newer Red Hat version (since Red Hat 5.0) whithin 1.5 weeks of release, but it took me around 2 months to step up to Fedora 2. Why? Because the nvidia drivers didn't work with the fedora 2 kernel and I have better things to do with my time that patching and rebuilding kernels. When nvidia released new drivers I upgraded within 1 week. If I were a new user, I wouldn't even have bothered.
Carlos Rodrigues
On Thursday 30 September 2004 18:20, Carlos Rodrigues wrote:
Also that "not supported" thing is crap. Binary modules are a fact of life and some care ("some" meaning "break it if you have to, but avoid it if you can") must be taken so that users can use them.
Only if you need 3D acceleration do you need the closed source nvidia drivers to use nvidia cards. The nv driver does work fairly well, even if it isn't the ultimate in performance.
The Linux kernel developers have made the decision to not support binary modules. It's not a Red Hat decision. The developers can't be expected to help debug a problem with a module for which they don't have source, nor can they be expected to debug a problem in a kernel which has a binary-only module loaded, since they can't really be sure the binary module isn't the source of the problem.
So, when the word 'supported' is used in relation to binary modules, it's not that they can't be used it's just that if you choose to use them then you need to go to the supplier of the module for support.
I personally do use the nvidia driver (courtesy of ATrpms) on my personal box at home, but I understand the risks of doing so.
On Thu, 2004-09-30 at 18:20, Carlos Rodrigues wrote:
[snip]
Also that "not supported" thing is crap. Binary modules are a fact of life and some care ("some" meaning "break it if you have to, but avoid it if you can") must be taken so that users can use them.
Personally, I think that *zero* care should be taken to make sure that users can use binary modules. The kernel developers have spoken, and it is not their concern. With good reason, too, that has been discussed ad-nauseum on this list as well as many others. And this coming from a VMware user who gets an occasional kernel panic when rebooting his system but never reports it because he doesn't get it on any of his non-VMware running systems. It's my choice, it's my problem. But on a more constructive side, never underestimate the power of the Free Software world. If I remember correctly, it used be like pulling teeth to get the SCSI card vendors to open up their specs so that their cards could be supported by Linux. Then Linux started taking off in the server room and it wasn't long before they were falling over each other to offer either specs, or complete (and GPLed) drivers to get their cards supported under Linux. Xircom (now owned by Intel) comes to mind, for one. Give it time...the video card vendors will one day, hopefully soon, be forced to open source their drivers, or give complete specs to implement drivers so that they can be first to be fully supported (included 3D acceleration) under Linux. Er...maybe I shouldn't say 'forced'. Rather, they will *also* be falling over each other to get their cards supported on stock GNU/Linux based operating systems.
-
Carlos Rodrigues -
Florin Malita -
Kyrre Ness Sjobak -
Lamar Owen -
Paul Iadonisi