On Wed, Jul 19, 2023 at 6:23 AM Jaroslav Mracek <jmracek(a)redhat.com> wrote:
> On Mon, Jul 17, 2023 at 6:40 AM Jaroslav Mracek <jmracek(a)redhat.com>
wrote:
>
> Except dnf5 broke a number of microdnf usecases with low memory where
> microdnf worked [1].
>
> [1]
https://bugzilla.redhat.com/show_bug.cgi?id=2214520
Correct but as you can see the issue was not in DNF5 but in libsolv (solver for DNF,
DNF5, zypper, PackageKit). Ales Matej from DNF team prepared two patches - one to resolve
the issue in libsolv and second to enable workaround for DNF5. Thanks to better DNF5
structure we were able to discover the real cause of the issue.
Does that mean the issues with dnf [2] we able to be solved all the
time but just weren't investigated?
Therefore I am curios why the issue is mentioned here?
Because the issues, while quite probably now known, still aren't
resolved in F-38 where it was meant to be stable for those usecases
from the release of F-38. This in turn gives me little confidence in
the rest of dnf5 being ready to replace the much larger set of
usecases as implemented by dnf4 by the Change Completion [3] deadline
in less than 3 weeks time.
[2]
https://bugzilla.redhat.com/show_bug.cgi?id=1907030
[3]
https://fedorapeople.org/groups/schedule/f-39/f-39-key-tasks.html