On Mon, Aug 3, 2009 at 20:26, Dan McGee wrote:

Holy hell, 454 ms for udev time. This is fast!

I'll signoff for x86_64 as long as my concerns about the kernel version requirements aren't swept under the rug. Not sure if it is news-item worthy or what the solution is.

-Dan

+1 to a news item and probably a post_upgrade as well

Am Sun, 23 Aug 2009 23:10:54 +1000 schrieb Allan McRae :

Jan de Groot wrote:

...

On Sun, 2009-08-02 at 21:08 +0200, Tobias Powalowski wrote:

...

Hi bump to latest udev version, please test it well. Has many new things included, which also dig in glib2 and libusb depend.

greetings tpowa

To bump this thread:

- udev 145 contains .la files, they should be removed - Allan has to patch glibc to support older kernels with signalfd

When this is done, we need to move: - e2fsprogs - util-linux-ng - udev - hal - initscripts - glibc

I don't know what's allan's schedule is for this, but I could release a glibc package with just that patch included. We have quite some stuck packages due to the old util-linux-ng and udev in core at this moment.

Please go ahead and release a glibc with that patch. I am quite busy at the moment so probably will not get to Arch stuff until later next week...

Allan

Hm. IMHO this would break our own patching rules though it may be helpful here. We usually don't apply patches for unsupported packages. And we currently only have one kernel to support in our repos. So I thinking that patch would not be allowed. I'm thinking about adding a long requested 2nd kernel back to our repos. It would be the wanted "fallback" when core kernel would fail booting after rebooting. I'd like to maintain myself a "longlife" upstream supported kernel that is right now kernelseries 2.6.27.xx with maybe all 3rd party modules + Xen support if possible. We should be able to support this 2nd kernel for a longer time with our udev and mkinitcpio until next longlife version will be ready. This one would become the recommended kernel for server usage (not sure if it should become a different preempt setting). We could have it either in core as 2nd one but extra would be also ok to me and would save us signoffs in case of quick security fixes. Not sure if it shoul be a choice for the iso. Opinions? -Andy

On Sun, Aug 23, 2009 at 9:12 AM, Andreas Radke wrote:

Am Sun, 23 Aug 2009 23:10:54 +1000 schrieb Allan McRae :

...

Jan de Groot wrote:

...

On Sun, 2009-08-02 at 21:08 +0200, Tobias Powalowski wrote:

...

Hi bump to latest udev version, please test it well. Has many new things included, which also dig in glib2 and libusb depend.

greetings tpowa

To bump this thread:

- udev 145 contains .la files, they should be removed - Allan has to patch glibc to support older kernels with signalfd

When this is done, we need to move: - e2fsprogs - util-linux-ng - udev - hal - initscripts - glibc

I don't know what's allan's schedule is for this, but I could release a glibc package with just that patch included. We have quite some stuck packages due to the old util-linux-ng and udev in core at this moment.

Please go ahead and release a glibc with that patch.  I am quite busy at the moment so probably will not get to Arch stuff until later next week...

Allan

Hm. IMHO this would break our own patching rules though it may be helpful here. We usually don't apply patches for unsupported packages. And we currently only have one kernel to support in our repos. So I thinking that patch would not be allowed.

This line of thinking is absurd. Patching rules? We had like 8 devs say this was a good thing, and the fucking patch is upstream! An excuse like "patching rules" here is ridiculous, and calling any non-stock kernel unsupported is also a great way to anger a large crowd of Arch users. You did approve this kernel patch, right? http://projects.archlinux.org/?p=linux-2.6-ARCH.git;a=blob;f=patches/quirks-... Obviously I'm being a bit sarcastic here- I don't think you actually approved this patch, but my point is that we already patch our kernel for some things that affect a small proportion of our users with that hardware, and yet a extremely scope-limited yet useful glibc patch draws more fire. In light of recent activity in the dev circle, I'd also like to ensure this isn't seen as a personal attack on you, Andy. I'm only frustrated that a crutch like "patching rules" could even be pulled out here. The second kernel idea is a good one. However, it still does not address the main reason a lot of us wanted this patch in, which is when you are running in an environment where you do *not* control the kernel. -Dan

On Sun, 2009-08-23 at 14:46 +0200, Jan de Groot wrote:

On Sun, 2009-08-02 at 21:08 +0200, Tobias Powalowski wrote:

...

Hi bump to latest udev version, please test it well. Has many new things included, which also dig in glib2 and libusb depend.

greetings tpowa

To bump this thread:

- udev 145 contains .la files, they should be removed - Allan has to patch glibc to support older kernels with signalfd

When this is done, we need to move: - e2fsprogs - util-linux-ng - udev - hal - initscripts - glibc

I don't know what's allan's schedule is for this, but I could release a glibc package with just that patch included. We have quite some stuck packages due to the old util-linux-ng and udev in core at this moment.

About udev 145: I had an early test version that I grabbed from a staging dir on my system. The official 145 package in testing doesn't have .la files. glibc has been patched for signalfd, so when initscripts, kernel-headers and glibc are signed off, this can go to core.