Hi,
usually ATrpms jumps into testing at test2 or test3. This time we'll try supporting FC6 from test1 onwards.
A couple of notes:
o ATrpms uses disttags of 5.90 matching the internal verisoning of test1. When test2 hits the road packages will all be rebuilt with 5.91 as a disttag and so on. That way upgrade paths are retained (but these are not guaranteed within test releases/rawhide anyway, so don't feel on the safe side) and rebuilds don't have to bump the buildtag in front of the disttag, so when your FC5 box has foo-1.2.3-4.* and test/rawhide bits also start with foo-1.2.3-4.* it means that the same sources/specfiles are still used.
o kernel support (kmdls) is there, both for the test release kernel, 2.6.16-1.2289_FC6, as well as hopefully the latest rawhide kernel, 2.6.17-1.2318_FC6. Oops, the latest is already 2.6.17-1.2328.fc6, and by the time you're reading this probably even further. The point is that ATrpms will try to support the latest rawhide kernel, but it certainly can't keep up to the pace of rawhide. :)
o Please report bugs, especially regression wrt FC5 on bugzilla.atrpms.net under fc6.
Thanks and have fun!
On Sat, 1 Jul 2006, Axel Thimm wrote:
Hi,
usually ATrpms jumps into testing at test2 or test3. This time we'll try supporting FC6 from test1 onwards.
A couple of notes:
o ATrpms uses disttags of 5.90 matching the internal verisoning of test1. When test2 hits the road packages will all be rebuilt with 5.91 as a disttag and so on. That way upgrade paths are retained (but these are not guaranteed within test releases/rawhide anyway, so don't feel on the safe side) and rebuilds don't have to bump the buildtag in front of the disttag, so when your FC5 box has foo-1.2.3-4.* and test/rawhide bits also start with foo-1.2.3-4.* it means that the same sources/specfiles are still used.
o kernel support (kmdls) is there, both for the test release kernel, 2.6.16-1.2289_FC6, as well as hopefully the latest rawhide kernel, 2.6.17-1.2318_FC6. Oops, the latest is already 2.6.17-1.2328.fc6, and by the time you're reading this probably even further. The point is that ATrpms will try to support the latest rawhide kernel, but it certainly can't keep up to the pace of rawhide. :)
o Please report bugs, especially regression wrt FC5 on bugzilla.atrpms.net under fc6.
Cool. The only problem I see is that the nVIDIA kernel driver is bjorken with xorg 7.1. (And will be until they create a Linux driver for the 9xxx series of drivers.)
Are there x86_64 builds?
On Sat, Jul 01, 2006 at 09:24:59AM -0700, alan wrote:
On Sat, 1 Jul 2006, Axel Thimm wrote:
Hi,
usually ATrpms jumps into testing at test2 or test3. This time we'll try supporting FC6 from test1 onwards.
A couple of notes:
o ATrpms uses disttags of 5.90 matching the internal verisoning of test1. When test2 hits the road packages will all be rebuilt with 5.91 as a disttag and so on. That way upgrade paths are retained (but these are not guaranteed within test releases/rawhide anyway, so don't feel on the safe side) and rebuilds don't have to bump the buildtag in front of the disttag, so when your FC5 box has foo-1.2.3-4.* and test/rawhide bits also start with foo-1.2.3-4.* it means that the same sources/specfiles are still used.
o kernel support (kmdls) is there, both for the test release kernel, 2.6.16-1.2289_FC6, as well as hopefully the latest rawhide kernel, 2.6.17-1.2318_FC6. Oops, the latest is already 2.6.17-1.2328.fc6, and by the time you're reading this probably even further. The point is that ATrpms will try to support the latest rawhide kernel, but it certainly can't keep up to the pace of rawhide. :)
o Please report bugs, especially regression wrt FC5 on bugzilla.atrpms.net under fc6.
Cool. The only problem I see is that the nVIDIA kernel driver is bjorken with xorg 7.1. (And will be until they create a Linux driver for the 9xxx series of drivers.)
Are there x86_64 builds?
Yes, but only for the FC6t1 kernel (2.6.16-1.2289_FC6). I saw that the build breaks for rawhide kernels, so I didn't make the nvidia builds public, yet.
On Sat, 1 Jul 2006, Axel Thimm wrote:
On Sat, Jul 01, 2006 at 09:24:59AM -0700, alan wrote:
On Sat, 1 Jul 2006, Axel Thimm wrote:
Hi,
usually ATrpms jumps into testing at test2 or test3. This time we'll try supporting FC6 from test1 onwards.
A couple of notes:
o ATrpms uses disttags of 5.90 matching the internal verisoning of test1. When test2 hits the road packages will all be rebuilt with 5.91 as a disttag and so on. That way upgrade paths are retained (but these are not guaranteed within test releases/rawhide anyway, so don't feel on the safe side) and rebuilds don't have to bump the buildtag in front of the disttag, so when your FC5 box has foo-1.2.3-4.* and test/rawhide bits also start with foo-1.2.3-4.* it means that the same sources/specfiles are still used.
o kernel support (kmdls) is there, both for the test release kernel, 2.6.16-1.2289_FC6, as well as hopefully the latest rawhide kernel, 2.6.17-1.2318_FC6. Oops, the latest is already 2.6.17-1.2328.fc6, and by the time you're reading this probably even further. The point is that ATrpms will try to support the latest rawhide kernel, but it certainly can't keep up to the pace of rawhide. :)
o Please report bugs, especially regression wrt FC5 on bugzilla.atrpms.net under fc6.
Cool. The only problem I see is that the nVIDIA kernel driver is bjorken with xorg 7.1. (And will be until they create a Linux driver for the 9xxx series of drivers.)
Are there x86_64 builds?
Yes, but only for the FC6t1 kernel (2.6.16-1.2289_FC6). I saw that the build breaks for rawhide kernels, so I didn't make the nvidia builds public, yet.
The problem with the nVIDIA kernel module is not the kernel. It is that the ABI for Xorg changed. The driver builds fine, but display is weird. Font display is iffy. There is a way to get it working, but it involved disabling render and abi.
It is supposed to be fixed in the next revision of the nVIDIA driver, but I have yet to get a timeline as to when that will be released. (And the new driver will be needed to get alglx running.)
On Sat, Jul 01, 2006 at 12:43:29PM -0700, alan wrote:
Cool. The only problem I see is that the nVIDIA kernel driver is bjorken with xorg 7.1. (And will be until they create a Linux driver for the 9xxx series of drivers.)
Are there x86_64 builds?
Yes, but only for the FC6t1 kernel (2.6.16-1.2289_FC6). I saw that the build breaks for rawhide kernels, so I didn't make the nvidia builds public, yet.
BTW the next ATrpms update will contain FC6 kmdls for rawhide kernels (and userland), too. The constraints you mention below are still valid, though.
The problem with the nVIDIA kernel module is not the kernel. It is that the ABI for Xorg changed. The driver builds fine, but display is weird. Font display is iffy. There is a way to get it working, but it involved disabling render and abi.
Yes, you need Option "RenderAccel" "0"
It is supposed to be fixed in the next revision of the nVIDIA driver, but I have yet to get a timeline as to when that will be released. (And the new driver will be needed to get alglx running.)
I guess currently in FC6 scope one should avoid the nvidia closed source driver. Hopefully there will soon be a fix. But I have the impression that nvidia will probably wait until FC6 reaches closer to release to fix any other new issues creeping up until then. Let's see.