On Sat, Aug 22, 2009 at 16:55, Kevin J. Cummings <<a href="mailto:cummings@kjchome.homeip.net">cummings@kjchome.homeip.net</a>> wrote: <div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> <div class="im">On 08/22/2009 04:00 PM, Ryan Lynch wrote: > On Sat, Aug 22, 2009 at 15:44, Kevin J. Cummings </div><div><div>> <<a href="mailto:cummings@kjchome.homeip.net">cummings@kjchome.homeip.net</a> <mailto:<a href="mailto:cummings@kjchome.homeip.net">cummings@kjchome.homeip.net</a>>> wrote: > > On 08/22/2009 03:39 PM, Geoffrey Leach wrote: > > On 08/22/2009 12:24:08 PM, Kevin J. Cummings wrote: > >> On 08/22/2009 03:17 PM, Ryan Lynch wrote: > >>> Running 'yum update' from the command line, I am getting some file > >>> conflicts between different arch/versions of the 'glibc' and > >>> 'glibc-common' packages. This machine runs F11-x86_64. > >>> > >>> Here's the actual error output: > >>> > >>> <code> > >>> Transaction Check Error: > >>> file /usr/share/doc/glibc-2.10.1/NEWS from install of > >>> glibc-2.10.1-4.i686 conflicts with file from package > >> glibc-2.10.1-2.x86_64 > >>> file /usr/bin/ldd from install of glibc-common-2.10.1-4.i586 > >> conflicts > >>> with file from package glibc-common-2.10.1-2.x86_64 > >>> file /usr/lib/locale/locale-archive.tmpl from install of > >>> glibc-common-2.10.1-4.i586 conflicts with file from package > >>> glibc-common-2.10.1-2.x86_64 > >>> file /usr/share/doc/glibc-common-2.10.1/ChangeLog.bz2 from > >> install > >> of > >>> glibc-common-2.10.1-4.i586 conflicts with file from package > >>> glibc-common-2.10.1-2.x86_64 > >>> file /usr/share/i18n/locales/de_AT from install of > >>> glibc-common-2.10.1-4.i586 conflicts with file from package > >>> glibc-common-2.10.1-2.x86_64 > >>> </code> > >>> > >>> Is this a known issue, or did I screw something up? > >> > >> Something is screwed up. Possibly on your system. glibc-common.i586 > >> is > >> conflicting with glibc-common.i686. I think you need one and not the > >> other. Since you system is a base x86_64, I would suggest trying the > >> following: > >> > >> # yum shell > >>> remove glibc-common.i586 > >>> install glibc-common.i686 > >>> run > >> > >> and see if that helps.... > > > > Hmmmm ... perhaps install glibc-common.x86_64? i686 is still 32 bit. > > Not if its a multi-lib problem. I assumed that the x86_64 part would > straighten itself out.... But yes, he needs to ensure that all of his > secondary libs are either all i586 or all i686 (where he has a choice > between the two) and that they co-exist well with their x86_64 > counterparts (which should be the primary concern anyways). > > > > Here's the output of `yum list installed glibc-common glibc`, for reference: > > <code> > Installed Packages > glibc.i686 2.10.1-2 installed > glibc.x86_64 2.10.1-2 installed > glibc-common.x86_64 2.10.1-2 installed > </code> > > I tried `yum remove glibc.i686`, but it wants to remove 116 packages, > for dependencies. Mostly, it looks like stuff I installed while getting > Skype for Linux to run, so I'm pretty sure its going to break some > things if I remove it all. </div></div></blockquote><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div><div class="h5"> </div></div>Ryan, You are barking up the wrong tree. Your problem is glibc-common.i586. Something is trying to install this package and it conflicts with your currently installed glibc-common.i686. You need to track down where that dependency is coming from and deal with it.</blockquote><div> No, I'm pretty sure that's the right tree :-) I didn't post the whole output of 'yum update', though, so that's not your fault. See below. </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div class="im"> > Isn't this a packaging bug? There's no conflict between my > currently-installed 'glibc.i686' and 'glib.x86_64' (2.10.1-2), although > there is a conflict between the two archs' 'glibc-common' packages. </div> Possibly, but we need to know what is causing it first. i586/i686 are not meant to be multilib (like x86_64/i686 is).</blockquote><div> I don't mean to be rude, but I think you have some misconceptions about how i686 and i586 work on Fedora, before we go any further: * There is no glibc-common.i686 package. It doesn't exist, either on my system or anywhere in stable Fedora. Go take a look at at the repos, right now, and confirm that. (There is such a package in Rawhide, but that's not relevant, here.) * Many source RPMs don't have an i686 binary package, at all. Instead, they have an i586 or i386 package for 32-bit. OpenSSH is like this (go check for yourself), and there are others. I don't exactly know why this is, but it's perfectly valid to have a mix of i586 and i686 packages on a Fedora system. * glibc-2.10.1-4.i686 is the package causing YUM to pull in glibc-common.i586. Here's the snippet from my original 'yum update' command: <output> ---> Package glibc.i686 0:2.10.1-4 set to be updated --> Processing Dependency: glibc-common = 2.10.1-4 for package: glibc-2.10.1-4.i686 </output> * glibc-2.10.1-2.i686 depends on 'glibc-common = 2.10.1-2', but it apparently tolerates either the i586 OR the x86_64 version of 'glibc-common'. The 2.10.1-4.i686 update, however, changed its dependencies: It only accepts the i586 arch of 'glibc-common'. You can confirm this by comparing: - `yum deplist glibc-2.10.1-2.i686 | grep glibc-common`: <output> dependency: glibc-common = 2.10.1-2 provider: glibc-common.x86_64 2.10.1-2 provider: glibc-common.i586 2.10.1-2 dependency: glibc-common = 2.10.1-2 provider: glibc-common.x86_64 2.10.1-2 provider: glibc-common.i586 2.10.1-2 </output> against: - `yum deplist glibc-2.10.1-4.i686 | grep glibc-common`: <output> dependency: glibc-common = 2.10.1-4 provider: glibc-common.i586 2.10.1-4 </output> IN SHORT: The package 'glibc.i686' changes its dependencies from the -2 release to -4. Previously, my 'glibc-common.x86_64' was sufficient to satisfy 'glibc.i686', whereas now, it *must* have 'glibc-common.i586'. All that make sense? I think at this point, I need to find out why that change happened, and whether there's anything to be done about it upstream. -Ryan </div></div>