Tag Info

New answers tagged

1

It's a cluster of RPM bugs. Not just one bug, or two bugs. A nest of the critters. RPM fails (failed?) to validate signed packages, didn't understand v4 GPG signatures but didn't notice it didn't understand them, didn't understand some key sizes and types but didn't notice it didn't understand that, and also choked on subkeys! This lifesaving blog entry by ...


1

You cannot install two different numbered versions of the i386 and x86_64 packages without causing a lot of breakage. Just install the same version of both packages. yum install glibc-devel.{i386,x86_64} After that, if you still receive this error, it's Oracle's fault and Oracle must fix it.


0

You can run yum groupinfo "GNOME Desktop Environment", which on an EL5 system, will display: Group: GNOME Desktop Environment Description: GNOME is a powerful, graphical user interface which includes a panel, desktop, system icons, and a graphical file manager. Mandatory Packages: control-center gnome-applets gnome-panel gnome-session ...


2

Stop the install command and rebuild your rpm database. You should be able to: rm /var/lib/rpm/__db* Then: rpm --rebuilddb Then attempt your installation again.


0

I used rpm2cpio to convert the only suitable .rpm file I found into a cpio archive on standard out. $ rpm2cpio php-mcrypt-5.4.16-1.el6.x86_64.rpm | cpio -idmv I then copied mcrypt.so an mcrypt.ini to the proper folders for the PHP 5.4 install. Voilá, mcrypt works now!


1

Unfortunately, the CentOS SCL collection for php54 does not include php54-mcrypt. You can take the SRPM for the php-mcrypt-5.4 and modify the spec to change the package name, install location, etc. In other others, create your own package (check rpmbuild). Or you could also use rcollet's repository as an alternative.


0

It's a bad idea to uninstall yum (specially because all those dependencies will trigger more and more packages that need to be removed), but that does not break the RPM database, which still contains all packages you have installed. If you need libxml 2.7 for a specific software, one option is to manually compile it and install in a different directory, ...


1

You can specify the unrolled directory with a -n parameter to %prep.


1

The problem is not Fedora 20, nor is it Mock. It's RHEL 5. The RPM macro %{_sharedstatedir} was set to %{prefix}/com in both RHEL 4 and RHEL 5. The problem is that the macro was never used to build those distributions, so nobody paid any attention to its value. It only became an issue when people began backporting packages from later Fedora which did use ...


0

The only way I have managed to do this is to select the servers throught the SSM in the gui and re-assign the channels through there rather than in a script


1

It would be tricky to bundle all the dependencies into a single RPM, not least because the system you build on may have a different package composition from your target (e.g., your target has newer, or conflicted packages already installed, which you may unintentionally overwrite). However, you could instead obtain a list of all prerequisite packages and ...


3

I think what you are looking for should be to statically compile the application. This will make it very unlikely that the user will get any dependencies that are not available for them already.


3

You don't need to bundle dependencies. If the end user doesn't have Internet access, that's their problem (they use the installation media to install and update packages).


0

Michael Hampton's comment is very relevant - Fedora 8 is ancient. It would be a very good idea to get yourself on something a little more modern! All that being said, it's entirely possible that both 5.2 and 5.3 are installed, but that 5.2 appears first in the path. If that's the case, it'll run 5.2 because it's the first one that it finds. You can find ...



Top 50 recent answers are included