Re: [arch-dev-public] [signoff] coreutils-7.5
On Tue, 2009-08-25 at 00:10 +1000, Allan McRae wrote:
upstream update remove autoconf patch (no longer needed) add utimensat patch to make compatible with kernels earlier than 2.6.22 build against and add libcap dep
Allan
Without that patch, what kernels does coreutils support then? Since udev breaks with pretty much everything below 2.6.24, I don't think it's useful to patch packages to support anything below 2.6.24.
Jan de Groot wrote:
On Tue, 2009-08-25 at 00:10 +1000, Allan McRae wrote:
upstream update remove autoconf patch (no longer needed) add utimensat patch to make compatible with kernels earlier than 2.6.22 build against and add libcap dep
Allan
Without that patch, what kernels does coreutils support then? Since udev breaks with pretty much everything below 2.6.24, I don't think it's useful to patch packages to support anything below 2.6.24.
It breaks support for kernels below 2.6.22. As our glibc supports 2.6.18 and above and people can have a static dev, so I thought this was a good patch to include. Anyway, it is an upstream patch, so the next release will make this moot... Allan
Allan McRae schrieb:
Without that patch, what kernels does coreutils support then? Since udev breaks with pretty much everything below 2.6.24, I don't think it's useful to patch packages to support anything below 2.6.24.
It breaks support for kernels below 2.6.22. As our glibc supports 2.6.18 and above and people can have a static dev, so I thought this was a good patch to include. Anyway, it is an upstream patch, so the next release will make this moot...
Hmmmm, http://projects.archlinux.org/?p=initscripts.git;a=commitdiff;h=76bb11cb8b53... I think this hasn't worked for years anyway.
Thomas Bächler wrote:
Allan McRae schrieb:
Without that patch, what kernels does coreutils support then? Since udev breaks with pretty much everything below 2.6.24, I don't think it's useful to patch packages to support anything below 2.6.24.
It breaks support for kernels below 2.6.22. As our glibc supports 2.6.18 and above and people can have a static dev, so I thought this was a good patch to include. Anyway, it is an upstream patch, so the next release will make this moot...
Hmmmm, http://projects.archlinux.org/?p=initscripts.git;a=commitdiff;h=76bb11cb8b53...
I think this hasn't worked for years anyway.
Uh... so we can not have a static dev fs? Wasn't that the whole argument against bumping the kernel version required for glibc too high? Or has the situation changed enough with the new udev to revisit that issue? Allan
Allan McRae schrieb:
Uh... so we can not have a static dev fs? Wasn't that the whole argument against bumping the kernel version required for glibc too high? Or has the situation changed enough with the new udev to revisit that issue?
Before http://projects.archlinux.org/?p=initscripts.git;a=commitdiff;h=623e3fac060b... it wasn't possible anyway, you just got an empty ramfs in /dev. Since then, you could theoretically have your static nodes in /lib/udev/devices, run pacman -Rd udev (initscripts depend on it) and they would be copied over to /dev on startup. I doubt anyone even tried to do that. It will still work, it will simply try to start udev, fail and use the static /dev. I just don't see a point in having a code path here which we don't support at all.
participants (3)
-
Allan McRae
-
Jan de Groot
-
Thomas Bächler