Re: [Cooker] apt for mandrake
On Thu, Jun 26, 2003 at 08:35:05PM +0200, Christiaan Welvaart wrote: On Wed, 25 Jun 2003, Thierry Vignaud wrote: however, it really is buggy I believe rpmdsCompare doesn't work very well and apt didn't work around it. That should be fixed now. I set up an apt source for mdk 9.1 distribution and updates. It is updated hourly, but is mainly avaliable internationally for testing. If it gets overused I will close it down for international traffic. It is available as: rpm http://klid.dk/sw/apt mandrake/9.1 os contrib updates in your /etc/apt/sources.list However when I tried to use it, an apt-get upgrade wanted to delete about 300 packages, so something seems to be wrong in the dependencies, or maybe in the APIs. I did all this on the mdk 9.1 apt package for the client side, redhat packages for the server side. Any suggestions for getting it to work? Best regards keld
Re: [Cooker] apt for mandrake
On Thu, Jun 26, 2003 at 08:35:05PM +0200, Christiaan Welvaart wrote: On Wed, 25 Jun 2003, Thierry Vignaud wrote: Christiaan Welvaart [EMAIL PROTECTED] writes: The following packages have unmet dependencies: util-linux: Depends: shadow-utils (= 2902-5) but 1:4.0.3-5mdk is installed apt bug: 1:1 is greater than 0: Of course, but assuming epoch=0 when epoch is missing from version does not work. For example gimp requires libgimp1.2 = 1.2.3-21mdk, but libgimp1.2 has version 1:1.2.3-21mdk. You can see all of these as warnings in the output ofrpm -Vav --nofiles I looked a bit at the version compare code but it seems ok to me, handling epochs and even using rpmlib to compare versions. however, it really is buggy I believe rpmdsCompare doesn't work very well and apt didn't work around it. That should be fixed now. The sources.list is a bit different than regular apt-rpm, the supplied default sources.list reflects that. i586: http://www.time4t.net/~cjw/mdk-cooker/RPMS/apt-0.5.5cnc6-2mdk.i586.rpm http://www.time4t.net/~cjw/mdk-cooker/RPMS/libapt0.5-0.5.5cnc6-2mdk.i586.rpm http://www.time4t.net/~cjw/mdk-cooker/RPMS/libapt0.5-devel-0.5.5cnc6-2mdk.i586.rpm http://www.time4t.net/~cjw/mdk-cooker/RPMS/synaptic-0.37-1mdk.i586.rpm ppc: http://www.time4t.net/~cjw/mdk-cooker-ppc/RPMS/apt-0.5.5cnc6-2mdk.ppc.rpm http://www.time4t.net/~cjw/mdk-cooker-ppc/RPMS/libapt0.5-0.5.5cnc6-2mdk.ppc.rpm http://www.time4t.net/~cjw/mdk-cooker-ppc/RPMS/libapt0.5-devel-0.5.5cnc6-2mdk.ppc.rpm http://www.time4t.net/~cjw/mdk-cooker-ppc/RPMS/synaptic-0.37-1mdk.ppc.rpm src: http://www.time4t.net/~cjw/mdk-cooker/SRPMS/apt-0.5.5cnc6-2mdk.src.rpm http://www.time4t.net/~cjw/mdk-cooker/SRPMS/synaptic-0.37-1mdk.src.rpm Hmm, tried to install things on my mdk 9.1 system, but i did not have librpm4.2 only librpm4.0.4 Would it be possible for you to build it with a dependency on librpm4.0 ? best regards keld
Re: [Cooker] apt for mandrake
On Wed, 25 Jun 2003, Thierry Vignaud wrote: Christiaan Welvaart [EMAIL PROTECTED] writes: The following packages have unmet dependencies: util-linux: Depends: shadow-utils (= 2902-5) but 1:4.0.3-5mdk is installed apt bug: 1:1 is greater than 0: Of course, but assuming epoch=0 when epoch is missing from version does not work. For example gimp requires libgimp1.2 = 1.2.3-21mdk, but libgimp1.2 has version 1:1.2.3-21mdk. You can see all of these as warnings in the output ofrpm -Vav --nofiles I looked a bit at the version compare code but it seems ok to me, handling epochs and even using rpmlib to compare versions. however, it really is buggy I believe rpmdsCompare doesn't work very well and apt didn't work around it. That should be fixed now. The sources.list is a bit different than regular apt-rpm, the supplied default sources.list reflects that. i586: http://www.time4t.net/~cjw/mdk-cooker/RPMS/apt-0.5.5cnc6-2mdk.i586.rpm http://www.time4t.net/~cjw/mdk-cooker/RPMS/libapt0.5-0.5.5cnc6-2mdk.i586.rpm http://www.time4t.net/~cjw/mdk-cooker/RPMS/libapt0.5-devel-0.5.5cnc6-2mdk.i586.rpm http://www.time4t.net/~cjw/mdk-cooker/RPMS/synaptic-0.37-1mdk.i586.rpm ppc: http://www.time4t.net/~cjw/mdk-cooker-ppc/RPMS/apt-0.5.5cnc6-2mdk.ppc.rpm http://www.time4t.net/~cjw/mdk-cooker-ppc/RPMS/libapt0.5-0.5.5cnc6-2mdk.ppc.rpm http://www.time4t.net/~cjw/mdk-cooker-ppc/RPMS/libapt0.5-devel-0.5.5cnc6-2mdk.ppc.rpm http://www.time4t.net/~cjw/mdk-cooker-ppc/RPMS/synaptic-0.37-1mdk.ppc.rpm src: http://www.time4t.net/~cjw/mdk-cooker/SRPMS/apt-0.5.5cnc6-2mdk.src.rpm http://www.time4t.net/~cjw/mdk-cooker/SRPMS/synaptic-0.37-1mdk.src.rpm Christiaan
[Cooker] apt for mandrake
For the people who want to check out synaptic on mandrake I tried to cook an apt that works with regular mandrake mirrors. It seems to work ok, but on my systems it reports a dependency problem. And since it's apt, it refuses to do anything as long as there are problems. Without this dep problem it happily (dist-)upgrades and installs cooker packages. The following packages have unmet dependencies: util-linux: Depends: shadow-utils (= 2902-5) but 1:4.0.3-5mdk is installed I looked a bit at the version compare code but it seems ok to me, handling epochs and even using rpmlib to compare versions. It can be worked around by building a util-linux that depends on a different version of shadow-utils, but that's not a solution. Does anyone have an idea what could be causing this dependency compare problem? If I rebuild the apt SRPM in contrib it reports the same, so maybe it's related to rpm 4.2. source rpm is at http://www.time4t.net/~cjw/mdk-cooker/SRPMS/apt-0.5.5cnc6-1mdk.src.rpm Christiaan
Re: [Cooker] apt for mandrake
Does anyone have an idea what could be causing this dependency compare problem? If I rebuild the apt SRPM in contrib it reports the same, so maybe it's related to rpm 4.2. no, it was the same when the 9.1 was out. So, this is a bug, either in apt or in rpmlib. you should ask the dev. -- Mickaël Scherer
Re: [Cooker] apt for mandrake
Christiaan Welvaart [EMAIL PROTECTED] writes: The following packages have unmet dependencies: util-linux: Depends: shadow-utils (= 2902-5) but 1:4.0.3-5mdk is installed apt bug: 1:1 is greater than 0: I looked a bit at the version compare code but it seems ok to me, handling epochs and even using rpmlib to compare versions. however, it really is buggy