On Saturday 03 January 2004 10:09, Shlomi Fish wrote: > Hetz: thanks for the info. As for the version of the driver, it may be > stable for you, but I'm not sure it would be stable for me. So I'll ask > at the forums, IRC channel, etc. > > The rest of the crowd: the Nvidia card came prepackaged with the computer, > and I can't talk my father into replacing without a good enough reason. > (he uses Windows there most of the time). So, switching to ATI is not an > option. > > The reason I believe the current situation with Nvidia cards is > sub-optimal is because: > > 1. I need to explictly download and build it whenever I upgrade the kernel > (and possibly X as well). Mandrake does not ship it with their distro so > they won't taint their distribution with a proprietary binary-only driver.
Yep. They try to make it user-friendly as much as possible. Closed, and yet, easy to install. You just run the installer, and it compiles what needs to be compiled for you. > > 2. It cannot be made part of the kernel because of its nature, so > upgrading a kernel is always a two step process. Upgrading a kernel is always more then two steps, anyhow. Besides checking that everything works. Upgrading NVidia's driver does not require re-configuration of X. > > 3. It causes some problems. Like this one, or one on my previous computer > where the X server completely freezed occasionally while the computer was > working. Why should it? A Linux machine should work flawlessly Usually the reason has to do with hardware competablility, and hardware inter-communication. Usually it's about the AGP. A tip (which solved a similar problem for me) - Add /etc/modules.conf: options agpgart agp_try_unsupported=1 It might solve the problem. It will avoid using NVidia's AGP driver, and use the kernel's AGP. > > 4. It "taints" the kernel and possibly make isolating problems a two-part > process (removing the driver and then testing the untainted kernel). > > So, Nvidia Corp. has done a nice gesture to the i386 Linux users, but > hasn't done enough. Linux "compatibility" is not enough. You still have to > play by the rules of open-source. Some do, some don't. Under the condition, they try to be as accessible as possible for the huge veriaty of distributions. Not many closed-source vendors are, or even bother trying... > > At the moment I don't have much time to try and reverse-engineer the > driver. (and I'm not sure what's the legal status of it). Even so, without > the SPEC, the re-created driver can still suffer from the same problems. > > Regards, > > Shlomi Fish > Ez. > > > > > ---------------------------------------------------------------------- > Shlomi Fish [EMAIL PROTECTED] > Home Page: http://t2.technion.ac.il/~shlomif/ > > Writing a BitKeeper replacement is probably easier at this point than > getting its license changed. > > Matt Mackall on OFTC.net #offtopic. > > > > ================================================================= > To unsubscribe, send mail to [EMAIL PROTECTED] with > the word "unsubscribe" in the message body, e.g., run the command > echo unsubscribe | mail [EMAIL PROTECTED] ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
