[criteria update] Package set

drago01 drago01 at gmail.com
Mon Sep 17 22:28:53 UTC 2012


On Tue, Sep 11, 2012 at 2:52 PM, Petr Schindler <pschindl at redhat.com> wrote:
> On Čt, 2012-09-06 at 09:57 -0700, Adam Williamson wrote:
>> On 2012-09-06 0:59, Kamil Paral wrote:
>> >> Hi,
>> >>
>> >> Because of changes in package set selection in new anaconda, I
>> >> propose
>> >> to amend the alpha criterion:
>> >>
>> >> 'The installer must be able to complete package installation with
>> >> the
>> >> default package set for each supported installation method'
>> >>
>> >> to:
>> >>
>> >> 'The installer must be able to install the default desktop for each
>> >> supported installation method (DVD, live, netinst, PXE, ...)'
>> >>
>> >> I chose default desktop because in f17 every installation method had
>> >> it
>> >> as default package set.
>> >
>> > It makes sense to adjust it, because there is no longer "default"
>> > package set. Also big thanks for clarifying what "installation
>> > methods" mean.
>>
>> Do we know if it's *intended* that there's no default package set, or
>> is that a bug? It only makes sense to amend the criterion if the lack of
>> a default package set is actually intended. Also, if it's intended that
>> there's no default package set, can there be said to be a 'default
>> desktop' any more? GNOME is only the 'default' in that it's the desktop
>> in the 'default package set'. If there's no 'default package set', GNOME
>> becomes simply a choice on the package set selection screen, co-equal
>> with all the others. I can't see how it can be called 'the default'.
>>
>> > I wonder - we require only the default desktop (GNOME) to be
>> > installable, but we have further Alpha criteria for other
>> > release-blocking desktops (KDE)? That's funny :-)
>>
>> > Maybe we should say:
>> > 'The installer must be able to (successfully) install *all
>> > release-blocking desktops* for each supported installation method
>> > (DVD, live, netinst, PXE, ...)'
>>
>> I think that's better, if we assume the new behaviour in anaconda is
>> actually intended.
>
> There is no default 'package set' now (by design, it's not a bug).

This *is* a bug IMO. We should have reasonable defaults and allow the
user to change them if he wants. But we should not force the user to
make choices that way. A user would have to know the differences
between the options to be able to make an informed decision. In case
the user has this knowledge he/she can open the spoke and change the
selection. In case the user doesn't it might more or less end up in a
random/wrong change.


More information about the test mailing list