Downgrade command ?

[Jan Engelhardt]

On Nov 26 2006 22:29, Pascal Bleser wrote:
>Jan Engelhardt wrote:
>> On Nov 26 2006 12:47, Mikus Grinbergs wrote:
>>
>> Speaking of 'experimental', here's my "don't-upgrade" request again:
>> 
>> If a channel is marked experimental, no packages are 
>> automatically selected for upgrade on `smart upgrade`, except for 
>> packages that are in rpm-sys.
>
>That's yet another behavior.
>It would mean tracking the origin of a package (the origin being the
>channel) and only upgrading a package from channel A if it was
>explicitly installed from channel A.

Yeah that's what I mean. It's a bit complicated to describe (and you did 
it right) and I'd just want it DTRT.

>OTOH, the problem is rather that package managers (including smart)
>don't track that information as it is not in the RPM database (which
>makes it problematic to track in the first place, as it is specific to
>the package manager). And they don't have any differentiation between
>packages coming from the distribution itself and 3rd party repositories.

$is_packman =~ /^[-\w]+-[\.\d]+-[\.\d+]\.pm\.\d+$/
$is_jengelh =~ /^[-\w]+-[\.\d]+-[\.\d+]\.jen\d+$/
$is_guru    =~ your funky naming style

well, so much about that. Maybe smart can operate on that. Or even the 
Packager: tag that rpm stores (rpm -qi) - yes that would be best!

>Point is, most people are working with exactly that situation every day,
>and have to make the decisions themselves.
>
>That goes far beyond my original requests about "downgrade" and "upgrade
>- --from" though ;)


	-`J'
--