yum-3.2.2-3.fc8 Using "yum upgrade", after the list is presented and y/n confirmation prompt, yum shows all the "---> Package .... to be updated" messages a second time.
fixed in upstream devel
yum-3.2.3-1.fc7 has been pushed to the Fedora 7 testing repository. If problems still persist, please make note of it in this bug report.
"Running rpm_check_debug" is still printed, but the duplicate messages are gone. It now prints things like: ts_done name in te is wine should be wine-jack
can you send me the output when you get the above messages and the yum.log of that transaction?
Created attachment 174381 [details] shell transcript Apparently relevant portion of /var/log/yum.log: Aug 27 13:48:01 Updated: wine-core - 0.9.43-2.fc7.i386 Aug 27 13:48:02 Updated: wine-cms - 0.9.43-2.fc7.i386 Aug 27 13:48:03 Updated: wine-tools - 0.9.43-2.fc7.i386 Aug 27 13:48:04 Updated: wine-capi - 0.9.43-2.fc7.i386 Aug 27 13:48:05 Updated: wine-esd - 0.9.43-2.fc7.i386 Aug 27 13:48:05 Updated: wine-nas - 0.9.43-2.fc7.i386 Aug 27 13:48:06 Updated: wine-twain - 0.9.43-2.fc7.i386 Aug 27 13:48:06 Updated: wine-jack - 0.9.43-2.fc7.i386 Aug 27 13:48:07 Updated: wine-ldap - 0.9.43-2.fc7.i386 Aug 27 13:48:13 Updated: wine-devel - 0.9.43-2.fc7.i386 Aug 27 13:48:13 Updated: wine - 0.9.43-2.fc7.i386 Aug 27 13:48:13 Updated: wine-docs - 0.9.43-1.fc7.noarch
cat transaction-all* install 0:wine-core-0.9.44-1.fc8.i386 install 0:wine-twain-0.9.44-1.fc8.i386 install 0:wine-tools-0.9.44-1.fc8.i386 install 0:wine-jack-0.9.44-1.fc8.i386 install 0:wine-capi-0.9.44-1.fc8.i386 install 0:wine-ldap-0.9.44-1.fc8.i386 install 0:wine-cms-0.9.44-1.fc8.i386 install 0:wine-nas-0.9.44-1.fc8.i386 install 0:wine-esd-0.9.44-1.fc8.i386 install 0:wine-0.9.44-1.fc8.i386 erase 0:wine-core-0.9.39-1.fc7.i386 erase 0:wine-twain-0.9.39-1.fc7.i386 erase 0:wine-tools-0.9.39-1.fc7.i386 erase 0:wine-jack-0.9.39-1.fc7.i386 erase 0:wine-capi-0.9.39-1.fc7.i386 erase 0:wine-ldap-0.9.39-1.fc7.i386 erase 0:wine-0.9.39-1.fc7.i386 erase 0:wine-cms-0.9.39-1.fc7.i386 erase 0:wine-nas-0.9.39-1.fc7.i386 erase 0:wine-esd-0.9.39-1.fc7.i386 when I start seeing that message this is what is in the transaction-all file in /var/lib/yum and the progress output on the screen doesn't mention wine being cleaned up. Running Transaction Updating : wine-core ####################### [ 1/20] Updating : wine-twain ####################### [ 2/20] Updating : wine-tools ####################### [ 3/20] Updating : wine-jack ####################### [ 4/20] Updating : wine-capi ####################### [ 5/20] Updating : wine-ldap ####################### [ 6/20] Updating : wine-cms ####################### [ 7/20] Updating : wine-nas ####################### [ 8/20] Updating : wine-esd ####################### [ 9/20] Updating : wine ####################### [10/20] Cleanup : wine-core ####################### [11/20] Cleanup : wine-twain ####################### [12/20] Cleanup : wine-tools ####################### [13/20] Cleanup : wine-jack ####################### [14/20] Cleanup : wine-capi ####################### [15/20] Cleanup : wine-ldap ####################### [16/20] Cleanup : wine-cms ####################### [17/20] ts_done name in te is wine should be wine-cms Cleanup : wine-nas ####################### [18/20] ts_done name in te is wine-cms should be wine-nas Cleanup : wine-esd ####################### [19/20] ts_done name in te is wine-nas should be wine-esd according to the transaction-all the next step before wine-cms is cleaned up is wine should be cleaned up. So something else is happening inside the transaction. I'm going to see if I can revert this set of pkgs and test some more.
and also worth noting - no 20/20 in that transaction call back. :)
okie doke - so far I can only make this happen for these pkgs. The fun part is this. when I setup the transaction to gather the transaction elements it is not the same as the transaction that runs, according to rpm. Asking panu to comment b/c while this might be a side effect of the callback changes to see the output like this I don't think this is actually a bug in yum. not yet ,at least. I could be persuaded though :)
Just a wild guess but strange behavior with the "wine" package might have to do with the fact that it's a metapackage with no files at all. If that's the case then it should be reproducable with "git" metapackage as well.
If rpm notices that a package has no files to remove, will it not perform the callback for the cleanup? It seems like it should do the callback no matter what, shouldn't it?
I was just making noises while looking through the sources to refresh my memory ;) The progress callbacks are normally called based on file information. For packages without any files the callback is "emulated" on install, but not on erase at the moment. That should be filed as a new bug, this is getting mixed up with all sorts of issues...
okay, I filed that bug. And I'm going to close this one b/c it's muddled and the original issue is resolved.