On 2020-05-13 04:59, Samuel Sieb wrote:
On 5/12/20 5:33 AM, Ed Greshko wrote:
> On 2020-05-12 20:18, Sreyan Chakravarty wrote:
>>
>> On 5/12/20 5:25 PM, Ed Greshko wrote:
>>> So, I feel you need to define what you mean by that word.
>>
>> I always prefer software that works as expected.
>
> Well, for all we know, the packagekit upgrade is expected to not remove
"running" kernel.
>
> As a matter of fact, after I wrote that, I bet that is the intended behavior. Why?
>
> Because in my "testing" I did do "pkcon upgrade-system 32" and
the upgrade from
> F30 to F32 was done *Live". If the process were to delete the current running
kernel it would
> have been a disaster.
dnf specifically will not remove the running kernel. Even if you are running the oldest
kernel, it will not be removed. It will remove the oldest kernel that you are not
running. So, this should be irrelevant to the packagekit issue.
Yes, "dnf" will not remove the running kernel. But, are we 100% certain that
"pkcon" follows the same
rule? I suppose I could try it on a VM to test. And even if it doesn't
"today", it didn't always use libdnf.
So, it could be a remnant.
--
The key to getting good answers is to ask good questions.