Custom Query (894 matches)

The EVR display during bcfg2's operation has an asterisk ('*') when there is no epoch present. The typical convention is to not display the epoch if it is None, sometimes even if it is 0.

Furthermore rpm does set noepoch=zeroepoch since several years now. The old noepoch < zeroepoch bug is from the promoteepoch ages that did quite a lot of adventurous loops when comparing (absent epoch of an installed file is set the same as one to be installed). It is safe and also required to assume that no epoch is the same as zero epoch.

On Wed, 2 Jul 2008 21:36:20 +0300 Axel Thimm wrote:

On Wed, Jul 02, 2008 at 01:14:13PM -0500, Narayan Desai wrote:

On Wed, 2 Jul 2008 19:20:26 +0300 Axel Thimm wrote:

Would it be possible ot add an attribute to the Ignore tag of what to ignore? E.g. in the multilib case I'd only like to ignore timestamps for some cases and not all attributes (like symlinks, owner, mode etc.).

Just so I am sure that I understand precisely what you are asking for, you want more fine-grained control of behavior during verification failures, so that you can describe single-file verification problems in a great deal of specificity? (ie don't turn off md5sum verification for the entire package, just do it for this individual file, and still flag permission problems on that file) Is this right?

Yes. :)

(it's all about the multilib issues that are 99% just different timestamps on config files or gzipped manpages)

Don't I know it ;)

I doubt that we'll get it into the next prerelease (targeted for really soon) but we should be able to get it done before too long. Can you file a ticket, so that we don't lose it?

When the [logging] path option is set in bcfg2.conf, the log that is produced has no timestamps. If the log was either written with timestamps or the client log was sent to the syslog daemon, the logs would then be available with timestamps, which would make them more useful.