| 1 |
Constrain _expand_new_virtuals output in order to avoid incorrect |
| 2 |
re-ordering of || deps in the dep_zapdeps function, as reported in |
| 3 |
bug 597752. |
| 4 |
|
| 5 |
The incorrect dep_zapdeps behavior involved a problem in the |
| 6 |
construction of the _dep_choice.cp_map dictionary inside the |
| 7 |
dep_zapdeps function, with this input: |
| 8 |
|
| 9 |
|| ( dev-java/icedtea-bin:8 =virtual/jdk-1.8.0-r3 >=virtual/jdk-1.5 ) |
| 10 |
( dev-java/icedtea-bin:7 =virtual/jdk-1.7.0-r2 >=virtual/jdk-1.5 ) |
| 11 |
|
| 12 |
The cp_map for virtual/jdk-1.7.0-r2 erroneously contained |
| 13 |
virtual/jdk-1.8.0-r3 because that was the highest virtual/jdk matched |
| 14 |
by the >=virtual/jdk-1.5 atom. This patch removes the >=virtual/jdk-1.5 |
| 15 |
atom from the _expand_new_virtuals output, in order to avoid triggering |
| 16 |
the erroneous dep_zapdeps behavior. The >=virtual/jdk-1.5 atom is not |
| 17 |
completely discarded, since the depgraph is able to access it via the |
| 18 |
virt_atom._orig_atom attribute. |
| 19 |
|
| 20 |
In order to demonstrate that this patch solves the problem, run the |
| 21 |
command `emerge ant-core` from inside a stage3 chroot, and see that |
| 22 |
virtual/jdk:1.8 is favored over virtual/jdk:1.7. |
| 23 |
|
| 24 |
X-Gentoo-Bug: 597752 |
| 25 |
X-Gentoo-Bug-URL: https://bugs.gentoo.org/show_bug.cgi?id=597752 |
| 26 |
--- |
| 27 |
pym/_emerge/depgraph.py | 6 +++++- |
| 28 |
pym/portage/dep/dep_check.py | 13 +++++++------ |
| 29 |
2 files changed, 12 insertions(+), 7 deletions(-) |
| 30 |
|
| 31 |
diff --git a/pym/_emerge/depgraph.py b/pym/_emerge/depgraph.py |
| 32 |
index 26037ad..9161914 100644 |
| 33 |
--- a/pym/_emerge/depgraph.py |
| 34 |
+++ b/pym/_emerge/depgraph.py |
| 35 |
@@ -4453,7 +4453,11 @@ class depgraph(object): |
| 36 |
# _UNREACHABLE_DEPTH for complete mode. |
| 37 |
virt_depth = parent.depth |
| 38 |
|
| 39 |
- chosen_atom_ids = frozenset(id(atom) for atom in mycheck[1]) |
| 40 |
+ chosen_atom_ids = frozenset(chain( |
| 41 |
+ (id(atom) for atom in mycheck[1]), |
| 42 |
+ (id(atom._orig_atom) for atom in mycheck[1] |
| 43 |
+ if hasattr(atom, '_orig_atom')), |
| 44 |
+ )) |
| 45 |
selected_atoms = OrderedDict() |
| 46 |
node_stack = [(parent, None, None)] |
| 47 |
traversed_nodes = set() |
| 48 |
diff --git a/pym/portage/dep/dep_check.py b/pym/portage/dep/dep_check.py |
| 49 |
index 9af4e65..9d2ca4b 100644 |
| 50 |
--- a/pym/portage/dep/dep_check.py |
| 51 |
+++ b/pym/portage/dep/dep_check.py |
| 52 |
@@ -188,13 +188,14 @@ def _expand_new_virtuals(mysplit, edebug, mydbapi, mysettings, myroot="/", |
| 53 |
raise ParseError("%s: %s '%s'" % \ |
| 54 |
(pkg, mycheck[1], depstring)) |
| 55 |
|
| 56 |
- # Pull in virt_atom which refers to the specific version |
| 57 |
- # of the virtual whose deps we're expanding. Also pull |
| 58 |
- # in the original input atom, so that callers can reliably |
| 59 |
- # check to see if a given input atom has been selected, |
| 60 |
- # as in depgraph._slot_operator_update_probe. |
| 61 |
+ # Replace the original atom "x" with "virt_atom" which refers |
| 62 |
+ # to the specific version of the virtual whose deps we're |
| 63 |
+ # expanding. The virt_atom._orig_atom attribute is used |
| 64 |
+ # by depgraph to map virt_atom back to the original atom. |
| 65 |
+ # We specifically exclude the original atom "x" from the |
| 66 |
+ # the expanded output here, since otherwise it could trigger |
| 67 |
+ # incorrect dep_zapdeps behavior (see bug #597752). |
| 68 |
mycheck[1].append(virt_atom) |
| 69 |
- mycheck[1].append(x) |
| 70 |
a.append(mycheck[1]) |
| 71 |
if atom_graph is not None: |
| 72 |
virt_atom_node = (virt_atom, id(virt_atom)) |
| 73 |
-- |
| 74 |
2.7.4 |
Archives