rawhide report: 20060121 changes

Rahul Sundaram sundaram at redhat.com
Tue Jan 24 18:25:52 UTC 2006


Nicolas Mailhot wrote:

>Le lundi 23 janvier 2006 à 10:23 -0800, Jesse Keating a écrit :
>  
>
>>On Sun, 2006-01-22 at 19:08 -0500, Neal Becker wrote:
>>    
>>
>>>I strongly believe that most users would prefer the smart behavior, or at
>>>least the option of being asked.  Specifically, if some dependencies have
>>>problems, offer to go ahead with other packages that don't have problems. 
>>>      
>>>
>>Working around broken deps is not a smart thing to automate.  While
>>there are hard deps, there maybe some soft deps that are just unknown.
>>When testing, tests are performed with ALL the updates in place, not a
>>smattering of them
>>    
>>
>
>This is all well and dandy for traditional base + updates systems, it's
>an assumption that's dead wrong for rolling releases like rawhide.
>
>This thread as shown nothing @rh checks rawhide iterations are
>self-consistent before pushing them. So there are no "good" distro
>states, only a string of "gray" system states, and it's totally wrong of
>yum to expect a "good" system state will appear some time in the future.
>
>Hell, in theory it would be possible for rawhide to never be in a state
>yum likes from FCx to FCx+1T1
>  
>
Just use a script like the one in 
http://fedoraproject.org/wiki/Tools/yum or write a yum plugin. Nobody is 
stopping a yum plugin from being written for specific use in rawhide. If 
it requires more that a yum plugin hash out the details in yum-devel 
list with the relevant developers.


-- 
Rahul 

Fedora Bug Triaging - http://fedoraproject.org/wiki/BugZappers




More information about the fedora-devel-list mailing list