On Thu, 2006-11-09 at 12:24 -0700, Kim Lux wrote: > On Thu, 2006-11-09 at 14:11 -0500, Kurt Wall wrote: > > > > > Can you imagine if we had to build drivers in this manner for EVERY > > > piece of hardware on our computers ? Should we be applauding Nvidia > > for > > > their approach to Linux OS support ? I think not ! > > > > Newsflash: that's how it used to work, if you had anything beyond > > a few select SCSI host adapters. > > I've been running Linux since 1.3.18, so I understand that. > > > > I'd love to know how an admin with 200 PCs with nvidia cards would > > > handle this. > > > > The admin wouldn't run testing code, would blindly update with > > "yum update" and, more generally, would not be as incompetent as you > > appear to be. That is, an admin with 200 PCs to support would not > > have > > gotten in the position you did in the first place. > > Granted he wouldn't run yum update. But he would upgrade kernels from > time to time. I guess he would build the nvidia module on one and > distribute it to each machine to be installed at the same time as the > kernel. Big pain. > Updating a farm of PCs is a big pain if it breaks. He would update one, test, tweak, test, tweak, test, etc. Then when ready would do a consistent update to all. Much less pain than breaking every one. Admins are always looking for the least painful way of moving forward. Your way of using the testing repo is not in their path. > -- > Kim Lux, Diesel Research Inc. > >
