Hide Forgot
Description of problem: expect-devel.i686 is available for x86_64 but expect.i386 is not. Is the omission on purpose or accidental? Version-Release number of selected component (if applicable): 5.44.1.15-2.el6 How reproducible: N/A Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
It's intentional. By default, -devel packages are pulled in for multilib and then we also pull in any other packages required for dependency resolution. None of the i686 packages in the RHEL 6 x86_64 tree specifically require expect.i686, so expect.i686 doesn't get pulled in.
expect-devel.i686 _does_ depend on expect.i686: /usr/lib/libexpect.so -> libexpect5.44.1.15.so There is no /usr/lib/libexpect5.44.1.15.so in any package. It _is_ in expect.i686 which is not available for x86_64. Please explain what the point of expect-devel.i686 is. It brings in nothing but a dead link if you install it. The fact that both expect-devel packages only depend on expect (without arch) only makes it install but doesn't help. On the other hand, everything becomes correct again if one installs expect.i686. It also pulls nothing in that isn't already there.
Description : Expect is a tcl application for automating and testing interactive applications such as telnet, ftp, passwd, fsck, rlogin, tip, etc. Expect makes it easy for a script to control another program and interact with it. This package contains development files for the expect library. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Moving to the 'expect' component. I can't speak specifically to whether expect-devel should Require expect or not. However, if that dependency does get added to the RPM, it would automatically pull expect.i686 into future x86_64 trees. Let's see what the package maintainer says.
Seems like a packaging bug, yes.
Yes, it's a packaging issue - the libexpect.so link should be in expect package, not in expect-devel.
One can always repackage -- for sure. Will you ever repackage in EL6?? (I hope not!) The plain *.so link is always in foo-devel. The obvious solution is to promote expect.i686 to the x86_64 repo. With that, eveythings falls in place. Nobody needs a new packaging scheme. Or are you headings for a WONTFIX?
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHBA-2012-0456.html
You simply removed biarch. Congratulations for this both simple and ingenious idea. Please post your account number here.