Strange package dependency problem

[Iain Rae]

seth vidal wrote:

>On Sun, 2004-03-21 at 16:20 +0000, Iain Rae wrote:
>
>  
>
>>seth vidal wrote:
>>
>>    
>>
>>>>If you're dealing with large numbers of systems then automagic updates 
>>>>(from a stable) are a necessity, however you have to back them up with 
>>>>reporting/monitoring systems. I wouldn't expect yum to do all that, it's 
>>>>too site specific. Having yum mail root a summary might be useful but 
>>>>it's fairly trivial to add that into the above cron job.
>>>>   
>>>>
>>>>        
>>>>
>>>Something to keep in mind - if you're dealing with large numbers of
>>>systems and you're updating them from rawhide then you're asking for
>>>pain in general, not just b/c of the updating.
>>> 
>>>
>>>      
>>>
>>At work we're updating from redhat (they're mostly redhat9  boxes), 
>>Fedora, freshrpms,,  vendor/project  supplied and homebrew rpms,  you 
>>can learn to cope with certain levels of pain :)
>>We use a staging repository so nothing goes on completely blindly, but 
>>the thought of updating hundreds of hosts by hand every night is about 
>>as painful as the thought of telling them all to ignore errors.
>>
>>    
>>
>
>you're updating to the development tree?
>  
>

sorry, not quite sure what you're asking:

Are we updating whole machines to rawhide? no.
Do we pull indivdual (s)rpms out of rawhide? a small number yes, but 
it's usually the last on the list before writing our own rpms from scratch. 
Do we feed packages/patches/reports back into rawhide? the policy is we 
should though It's left up to individual admins to do it for the 
packages they're responsible for.