Loïc Minier a écrit :
Hi,
On Tue, Feb 28, 2006, Filipus Klutiero wrote:
could you please clarify exactly which dependency relation is not being
handled correctly by synaptic? I would have guessed a recommendation of
gstreamer0.10-alsa by rhythmbox, but rhythmbox does not recommend
gstreamer0.10-alsa.
rhythmbox depends on gstreamer0.10-plugins-good which recommends
gstreamer0.10-alsa.
OK, thanks.
Also, synaptic not dealing with Recommends by default does not make
Recommends useless, since the users can look at the package dependencies
and see which recommendations may be missing, just like with dselect.
Consequently, it seems that this bug would be more appropriate as
"wishlist".
What you describe is "Suggests:", not "Recommends:". Please check
Policy 7.2 which is *extremely* clear on the importance of
"Recommends:":
Recommends
This declares a strong, but not absolute, dependency.
The Recommends field should list packages that would be found
together with this one in all but unusual installations.
I believe synaptic mimics an apt-get bug here.
(BTW, dselect does install Recommends by default.)
Cheers,
I was actually describing both Suggests and Recommends. Synaptic handles
both in the same way, which is not at all by default. I don't think
synaptic is affected by an apt-get bug. apt-get's behavior of not
installing recommendations by default is maybe poor, but apt-get does
show the recommendations without problems AFAIK, so synaptic should
really have no problem in changing the default behaviour, as long as a
dialog showing the recommendations checked and suggestions unchecked and
allowing to check and uncheck them is created.
Note that I disagree about policy being "*extremely* clear" about the
importance of Recommends. Policy's definition relies on two relative
terms, "strong" and "unusual". The only precise definition I can deduce
from this is that a recommended package should be installed together
with the package that recommends it in most cases. This means that
synaptic *should* install the recommendations by default, but it's not a
bug that it's not implemented (or cleanly implemented) yet. My answer
was mainly triggered by the fact that this bug has an important
severity, but as you didn't set the bug's severity after reassigning to
synaptic, this might not have been your intention, so I might be arguing
for nothing. Anyway, the bug is downgraded now.
