On Thu, Oct 20, 2005 at 04:36:59PM +0200, Sven Luther wrote: > On Sat, Oct 15, 2005 at 08:11:40PM +0200, Bastian Blank wrote: > > On Sat, Oct 15, 2005 at 07:43:02PM +0200, Sven Luther wrote: > > > In the light of this, i would indeed vote to move our package to > > > 2.6.14-rc4 > > > (versioned at 2.6.13.99.rc4 or something, and maybe with 2.6.14-0 abi > > > name), > > > > 2.6.13+2.6.14-r4 or 2.6.14~rc4, I don't know if the later is now > > supported by dak. Abiname is 2.6.14-rc4 without additional suffix. > > Hi Bastian, ... > > We were going to aim for a 2.6.14-rc4 release tomorrow, a bit in advance of > schedule, but the current package seems to be in better shape than 2.6.13-1 > previously available (it built on amd64 and i385, and now we have powerpc and > sparc in addition i hear). > > Anyway, horms did some fixes i didn't fully understand, and you undid it, and > since you are kind of the expert on that, i wonder if you are planing to fix > it the right way or not, and if you could provide us with a bit of information > about what the problem is ?
My fixes are pretty easy to understsand. Without them the -rc4 gets interprated as an ABI name, and late in the build kernel-doc will fail, because kernel-package wants it to be 2.6.14-rc4, as per version.h, but control listgs 2.6.14. Well, that and every package is 2.6.14 abi rc4, instead of 2.6.14-4, abi 1. There is seconday problem, that apply needs to be taught how to deal with 2.6.13+2.6.14-rc4, vs 2.6.14-rc4, and it needs to be told that 2.6.14-rc4 is upstream, becuse it can't tell that 2.6.14-rc4 isn't 2.6.14 with a debnumber of rc4. In sort, run dpkg-buildpackage -rfakeroot -us -uc and the build will fail towards the end. If you want to use the original code, I'd suggest changing the version number to 2.6.14-1. I doubt anything more sophisticated will work. Incidently, the ABI can be changed in debian/arch/ I'd like to understand the need to overload the verison number even more than it is already. -- Horms -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
