Windows product version is used to check for MSI upgrades, it has the for $major.$minor.$build. Since virt-viewer has a different version $major.$minor.$micro, we are fitting $micro + $release in $build with this formula: $build = $micro * 256 + $release. However, -z releases version do not fit, making update 3.3z updates hard to push if the following rhevm release didn't rebase virt-viewer version. We need to change the formula, Daniel agreed that it could be something like: $build = $micro << 12 + ($release[0] << 4 + $release[1])
To simplify a bit, -z releases version upgrade are quite broken. In general, customer will have to perform uninstall first to make sure to use the next release...
$build = $micro << 12 + ($release[0] << 4 + $release[1]) With this scheme, we have limitation of: - 16 max micro version numbers - 256 release - 16 -z releases Imho, a better way would be to simplify upstream version scheme. 0.5.6 could be v5, 0.6.0 could nicely become v6. In other word, there are no clear meaning currently for the version scheme, but it cases a lot of trouble for windows version.
Adding Daniel in Cc
patch for configure.ac: https://www.redhat.com/archives/virt-tools-list/2014-June/msg00013.html With that, the spec will need to be updated to compute buildid=(%release[0] << 4 %release[1])
Posted a patch upstream to reduce virt-viewer versions to just major.minor instead of major.minor.micro, and bump version to 1.0. This will make it easy for us to translate z-stream release numbers into the $buildid field once RHEL rebases to 1.0 release. https://www.redhat.com/archives/virt-tools-list/2014-June/msg00025.html For our existing RHEL builds using older version scheme (ie 0.5.6 / 0.6.0) we'd have to do some short-term RHEL specific hack to munge $buildid in some manner.
(In reply to Marc-Andre Lureau from comment #1) > To simplify a bit, -z releases version upgrade are quite broken. In general, > customer will have to perform uninstall first to make sure to use the next > release... Hi Marc, For future bugs that does not require doc text, pls set the 'requres_doc_text' flag to '-'. I have removed the doc text for this bug and set the flag to -. Cheers, Julie
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. https://rhn.redhat.com/errata/RHSA-2015-0197.html