https://bugzilla.novell.com/show_bug.cgi?id=368209
User matz@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=368209#c6
--- Comment #6 from Michael Matz 2008-03-07 11:37:37 MST ---
FWIW the machine on which this happens is weierstrass. It still contains
the logfiles (/var/log/zypper.log). It shows that the input to the solver
inside zypper is different than the one inside deptestomatic. From
the deptestomatic log:
added 693 rpm rules because of weak dependencies
4149 of 19786 installable solvables considered for solving (rules has been
generated for)
pruned rules from 47353 to 36700
binary: 4476
normal: 32223, 84126 literals
decisions so far: 2
while from zypper.log:
added 818 rpm rules because of weak dependencies
4149 of 19786 installable solvables considered for solving (rules has been
generated for)
pruned rules from 47648 to 36759
binary: 4476
normal: 32282, 84236 literals
decisions so far: 2
So, it's different numbers of rules. I think that's because of namespace:
deps.
What is perhaps interesting for this machine and testcase is, that is has
several packages installed as both i586 _and_ x86_64 versions, e.g. libacl:
% rpm -q --qf="%{name}-%{version} %{arch}\n" -a 'libacl*'
libacl-2.2.45 i586
libacl-2.2.34 x86_64
The package update libzypp generates now creates two jobs:
<install channel="stable-x86" kind="package" name="libacl" arch="i586"
version="2.2.45" release="24" status="U_Tu_"/>
<install channel="stable-x86" kind="package" name="libacl" arch="x86_64"
version="2.2.45" release="24" status="U_Tu_"/>
But satsolver doesn't seem to like this, as it internally already created
a conflict rule:
Rule #7837:
!libacl-2.2.45-24.x86_64 [14442] (w1) Install.level1
!libacl-2.2.45-24.i586 [4675] (w2) Install.level1
next
rules: 7838 21411
JOB Rule
#37188:
libacl-2.2.45-24.x86_64 [14442] (w1) Install.level1
next
rules: 0 0
JOB Rule
#36704:
libacl-2.2.45-24.i586 [4675] (w1) Install.level1
next
rules: 0 0
This isn't really correct. You _can_ install both rpms, they create
no conflict. Hence also both jobs rules are okay. As jobrules are
singletons, and the assert is about singletons perhaps that has something
to do with it.
--
Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.