Re: Preparing 2.6.15-rc7 for experimental
Moin, On Fri, Dec 30, 2005 at 09:49:50AM +0100, Sven Luther wrote: On Thu, Dec 29, 2005 at 11:28:59PM +0100, Frederik Schueler wrote: Hello, I would like to get on with the preparations of 2.6.15-rc7 and schedule an upload to experimental for tomorrow evening UTC. m68k is unable to keep up with that schedule, partly because I went on vacation, and partly because my upstream is still at 2.6.14 and I am not going to duplicate the work that is going to happen eventually in linux-m68k CVS, if somebody else wants to do that, fine with me. I tried to hack up something quick, but I am either getting a diff that does not apply, or a huge diff, that does not compile. This will have to wait until our CVS is at 2.6.15, which usually never happens before the new version is released. Even when that is done, I need to work on the config. The last version I tried on my Amiga (2.6.12) did not quite boot, since it did not load the SCSI drivers, it did detect the IDE devices though. I assume both SCSI and IDE drivers are built as modules and live in the initrd? Maybe this is only a minor problem on m68k then. Oh, and yes, this means the kernel now build-depend on ocaml-interp Oh, was this really necessary? 32MB wasted for ocaml and friends. Does ocaml even work on m68k? I only remember having lots of problems with it... Christian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Preparing 2.6.15-rc7 for experimental
On Mon, Jan 02, 2006 at 10:20:15PM +0100, Christian T. Steigies wrote: Moin, On Fri, Dec 30, 2005 at 09:49:50AM +0100, Sven Luther wrote: On Thu, Dec 29, 2005 at 11:28:59PM +0100, Frederik Schueler wrote: Hello, I would like to get on with the preparations of 2.6.15-rc7 and schedule an upload to experimental for tomorrow evening UTC. m68k is unable to keep up with that schedule, partly because I went on vacation, and partly because my upstream is still at 2.6.14 and I am not I know, and the idea is to have all other arches ready, and then have m68k come in in -2 or something, same as we did with 2.6.14, which i believe was fine with you. going to duplicate the work that is going to happen eventually in linux-m68k CVS, if somebody else wants to do that, fine with me. I tried to hack up something quick, but I am either getting a diff that does not apply, or a huge diff, that does not compile. This will have to wait until our CVS is at 2.6.15, which usually never happens before the new version is released. I know, which is why we make allowance for m68k :) Even when that is done, I need to work on the config. The last version I tried on my Amiga (2.6.12) did not quite boot, since it did not load the SCSI drivers, it did detect the IDE devices though. I assume both SCSI and IDE drivers are built as modules and live in the initrd? Maybe this is only a minor problem on m68k then. Ah, working on the configs is important indeed, but you need t owork with yaird initramfs-tools also. This is important, because we want to Oh, and yes, this means the kernel now build-depend on ocaml-interp Oh, was this really necessary? 32MB wasted for ocaml and friends. Does ocaml This is only ocaml-interp, which depends on ocaml-base-nox, and that's it, it will represent at worst around 1MB only, and ... even work on m68k? I only remember having lots of problems with it... ... if i remember your problem well, they would not have happened in this case, as they affected the full blown ocaml packages, and they should be fixed anyway. If it was necessary, well, it never is, but i am going to implement over time a full blown split-config that actually works as it should, and i prefer to do it in a language i master well, and which makes it easy to do this (handle easily list and trees, ocaml is the best suited for that). Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Preparing 2.6.15-rc7 for experimental
* Frederik Schueler wrote: The following architectures need their configs still to be updated: alpha Current svn (and rc7.experimental1 too) builds fine on alpha. Norbert -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Preparing 2.6.15-rc7 for experimental
On Thu, Dec 29, 2005 at 11:28:59PM +0100, Frederik Schueler wrote: Hello, I would like to get on with the preparations of 2.6.15-rc7 and schedule an upload to experimental for tomorrow evening UTC. Notice that there was a change in the way 2.6.15 Kconfig works, apparently, and now later occuring entries in the config file are ignored and only the first mention is used. I modified the way the config files are generated to no more simply catting together the snipplet, but to do smarter handling of the snipplet, reading in each of the options and keeping only the more specific, so make sure you have the latest version of the package (r5144). Notice that this does make Jonas's work on the cleanup script unecessary, altough it is compatible with any such cleaned up config snipplets. It is still necessary to keep multi-choice options like the memory model ones from being duplicates, and the next step of this work is to provide a split-config tool which can handle those the right way. Oh, and yes, this means the kernel now build-depend on ocaml-interp, and the script is written in ocaml, but hey, if i can learn perl, ruby or python, i guess others can learn ocaml too, which is the superior language anyway :) Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Preparing 2.6.15-rc7 for experimental
On Fri, Dec 30, 2005 at 09:49:50AM +0100, Sven Luther wrote: Notice that there was a change in the way 2.6.15 Kconfig works, apparently, and now later occuring entries in the config file are ignored and only the first mention is used. Just reverse the sorting of the files should fix that. Bastian -- Women are more easily and more deeply terrified ... generating more sheer horror than the male of the species. -- Spock, Wolf in the Fold, stardate 3615.4
Re: Preparing 2.6.15-rc7 for experimental
On Fri, Dec 30, 2005 at 02:22:29PM +0100, Bastian Blank wrote: On Fri, Dec 30, 2005 at 09:49:50AM +0100, Sven Luther wrote: Notice that there was a change in the way 2.6.15 Kconfig works, apparently, and now later occuring entries in the config file are ignored and only the first mention is used. Just reverse the sorting of the files should fix that. Ah, indeed, didn't think of that, anyway, the issue is solved now. Also, Jonas mentioned that Kconfig may see # foo is not set options differently than module or builtin ones, but i have some doubts about this at second thought. Anyway, with my method we guarantee we do the right thing, and i hope to evolve the tool to something more complete, and replacing split-config in the end. Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Preparing 2.6.15-rc7 for experimental
Hello, I would like to get on with the preparations of 2.6.15-rc7 and schedule an upload to experimental for tomorrow evening UTC. The following architectures need their configs still to be updated: alpha hppa m68k s390 Anyone objecting against this schedule? It still would be great to have 2.6.15-final uploaded and working on all architectures the same day upstream releases. Best regards Frederik Schueler -- ENOSIG signature.asc Description: Digital signature
Re: Preparing 2.6.15-rc7 for experimental
On Thu, Dec 29, 2005 at 11:28:59PM +0100, Frederik Schueler wrote: The following architectures need their configs still to be updated: hppa I can update hppa's... Could someone remind me how one goes about doing it? Cheers, Kyle -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Preparing 2.6.15-rc7 for experimental
Hello,
On Thu, Dec 29, 2005 at 06:20:56PM -0500, Kyle McMartin wrote:
I can update hppa's... Could someone remind me how one goes about
doing it?
you need split-config from trunk/scripts/ which probably won't work out
of the box because of the debian architecture name and the kernel one
mismatching:
In line 68, just replace ARCH=#{OPTIONS['arch']} with ARCH=parisc and it
should work.
If someone feels less ruby-illiterate than me, and would like to fix this
issue: the same problem occurs on amd64 (x86_64), powerpc (ppc) and sparc
(sparc64); we probably just need a simple arch mapping table for it.
If you never worked with split-config:
after you have run a first dpkg-buildpackage -B, which will stop
somewhere during an oldconfig run at the first unset option, go to
debian/build/build-hppa-none-parisc
and run
split-config -c oldconfig ../../arch/ flavour
after the oldconfig run is done, it will ask you where to put the
options; You will know best if the single options are per-flavour
or per-arch ones; there should not be any unassigned global option left
in rc7.
split-config is a bit tricky at the beginning, if you have any trouble
just ask :)
Best regards
Frederik Schueler
--
ENOSIG
signature.asc
Description: Digital signature
