[PlanetCCRMA] Weird Synaptic Errors

jim branagan ccrma@themusenashville.com
Wed Jan 12 22:50:02 2005


Almost all the time, when I try to install or upgrade multiple packages
with synaptic (particularly 'upgrade all'), I get any where from 1 to 50 
error messages of the form 'you have held broken packages', with no indication which packages are causing the problem.  Synaptic shows 0 broken packages.  The 'broken package' filter shows nothing (blank panel).

My first approach to bypassing this problem was to install/upgrade only one package (and its dependencies sometimes) at a time.  This always worked.

My second attempt was to split the upgrade into 2 sections, and execute both halves separately.  This also always works.

My third approach was to delete one 'suspicious' package from the programmed changes, and install/upgrade it later.  This also always works.

Now, I found out I don't need to do anything??!  Hitting the execute button a second time, with absolutely no changes to the programmed changes list also always works. 

Anyone else experiencing behavior like this?  I have an (virtually) up-to-date FC1/Planet CCRMA system, with the FC1, PlanetCCRMA and livna repositories enabled, and practically everything installed.  In particular, rpm, apt, and synaptic are the latest versions available right now.  Is something probably wrong with my setup, or is this normal behavior?  I don't see any changes (additions or deletions) in the programmed changes lists after the first execute failed.

It's no longer causing me any problems, since the second execute always works, but it does makes me a little nervous.

The second problem I'm seeing is that in synaptic, 'update list' always fails (with a message box that vanishes on its own in a fraction of a second - I have no idea what it says).  'apt-get update' always works though.  This only seems to be a problem on systems where I have unpacked the planet iso images onto my hard drive and use 'file:' instead of 'cdrom:' in sources.list.  Is this a known bug in synaptic?
Again, I have an acceptable work-around, so this isn't causing any trouble.

-- 
e. james branagan
the MUSE  Nashville, TN