Wow. That code used goto statements.
How crass.
Printable View
Wow. That code used goto statements.
How crass.
So does the linux kernel, and for good reasons: http://eradicus.blogsome.com/2009/10...to-statements/
Linus is awful.
He probably puts the braces on the wrong lines too. *sad panda*
How dare you insult the One True Coding Style :P
But the FSF are happy with closed code that is burnt onto a ROM (note their advice to the openmoko project to do that). Now if the Nvidia drive came in a ROM on the card, then the FSF would be happy to use it, but any problems like this would be unfixable.
personally i happily use the opensource drivers for my intel and amd cards.
yes in the openmoko case it would have been about 1MB, judging by the debian packages size. ( https://www.fsf.org/blogs/community/task2-openmoko ). I wonder if they would accept NAND chip if the ability to wrote to it was removed for example by not connecting some pins to the circuit board. putting 1GB of NAND on a graphics card would not do much to the price.
Why do people get upset because of this? It supports the "Freeness" cause. Information needs to be free. The NVidia blob makes sure that everybody can be root and therefore gain access to all information.
This is truly "Free" as in "Libre" rather than "beer". What the hell do you people want?
There is actually a huge difference between that and blob drivers. The problem with blob drivers is that they offer the potential for kernel compromise. With a bad chunk of firmware, the firmware itself is restricted to the device on which it is loaded. The kernel still needs some kind of driver to interface with the hardware/blobfirmware, hence that kernel driver protects the kernel from the bad blob firmware.
At least to some degree.
Think of it like this;
The chip itself on the device is some kind of undocumented magic box. What difference does it make if the magic box is entirely physical hardware and no part programmed firmware? You still don't know what its doing.