upgrading f11 -> rawhide: verifying the general technique for debugging

[Seth Vidal]

On Mon, 9 Nov 2009, Robert P. J. Day wrote:

> On Mon, 9 Nov 2009, Adam Williamson wrote:
>
>> On Mon, 2009-11-09 at 14:24 -0500, Seth Vidal wrote:
>>
>>> I would interpret it as impossible to decipher w/o the complete
>>> output. Please file a bug about it and include the entire output -
>>> don't snip anything out.
>>
>> I think it's just that only trying to upgrade udev - as he's doing -
>> doesn't cause a newer bluez to be pulled in. I don't see why it
>> should, it doesn't make any sense for udev to have a dependency on
>> bluez. I don't think we exactly support 'yum upgrade
>> $SOME_REALLY_IMPORTANT_PACKAGE' from a Rawhide repo when you're
>> running a stable release. Even less so than we support 'yum
>> upgrade'.
>
>  i'm tempted to agree, except that the error clearly shows that the
> new udev conflicts with an existing bluez file, so obviously there's
> *some* kind of inter-relationship there that doesn't seem to exist
> with the *newer* bluez packages.
>
>  in any event, i just did
>
>  # yum upgrade bluez\*
>
> and that did the trick -- i'm now in the midst of a 300+ package
> upgrade.  but i BZed this anyway:
>
>  https://bugzilla.redhat.com/show_bug.cgi?id=533925
>
> someone else can decide if, somehow, udev and bluez need to learn how
> to play nice.

the problem is this.

we can't tell if there is a file conflict until AFTER the dep resolution 
is done and the pkgs are downloaded. This is b/c we don't have the 
checksums of all the files in the metadata. And I'd bet we don't want it, 
either. It'd make everything even more hugerrific.

-sv