merge: forcefully mark files that we get from the second parent as dirty

After a hg merge, we want to include in the commit all the files that we
got from the second parent, so that we have the correct file-level
history.  To make them visible to hg commit, we try to mark them as dirty.

Unfortunately, right now we can't really mark them as dirty[1] - the
best we can do is to mark them as needing a full comparison of their
contents, but they will still be considered clean if they happen to be
identical to the version in the first parent.

This changeset extends the dirstate format in a compatible way, so that
we can mark a file as dirty:

Right now we use a negative file size to indicate we don't have valid
stat data for this entry.  In practice, this size is always -1.

This patch uses -2 to indicate that the entry is dirty.  Older versions
of hg won't choke on this dirstate, but they may happily mark the file
as clean after a full comparison, destroying all of our hard work.

The patch adds a dirstate.normallookup method with the semantics of the
current normaldirty, and changes normaldirty to forcefully mark the
entry as dirty.

This should fix issue522.

[1] - well, we could put them in state 'm', but that state has a
different meaning.

hgext/mq.py

@@ -1050,7 +1050,7 @@ class queue:
                 for f in m:
                     repo.dirstate.normal(f)
                 for f in mm:
-                    repo.dirstate.normaldirty(f)
+                    repo.dirstate.normallookup(f)
                 for f in forget:
                     repo.dirstate.forget(f)
 

mercurial/dirstate.py

@@ -202,18 +202,24 @@ class dirstate(object):
         self._incpath(f)
 
     def normal(self, f):
-        'mark a file normal'
+        'mark a file normal and clean'
         self._dirty = True
         s = os.lstat(self._join(f))
         self._map[f] = ('n', s.st_mode, s.st_size, s.st_mtime)
         if self._copymap.has_key(f):
             del self._copymap[f]
 
-    def normaldirty(self, f):
+    def normallookup(self, f):
         'mark a file normal, but possibly dirty'
         self._dirty = True
-        s = os.lstat(self._join(f))
-        self._map[f] = ('n', s.st_mode, -1, -1)
+        self._map[f] = ('n', 0, -1, -1)
+        if f in self._copymap:
+            del self._copymap[f]
+
+    def normaldirty(self, f):
+        'mark a file normal, but dirty'
+        self._dirty = True
+        self._map[f] = ('n', 0, -2, -1)
         if f in self._copymap:
             del self._copymap[f]
 
@@ -523,6 +529,7 @@ class dirstate(object):
                     st = lstat(_join(fn))
                 if (size >= 0 and (size != st.st_size
                                    or (mode ^ st.st_mode) & 0100)
+                    or size == -2
                     or fn in self._copymap):
                     madd(fn)
                 elif time != int(st.st_mtime):

mercurial/localrepo.py

@@ -998,7 +998,7 @@ class localrepository(repo.repository):
                 elif self.dirstate[f] in 'amn':
                     self.ui.warn(_("%s already tracked!\n") % f)
                 elif self.dirstate[f] == 'r':
-                    self.dirstate.normaldirty(f)
+                    self.dirstate.normallookup(f)
                 else:
                     self.dirstate.add(f)
         finally:

mercurial/merge.py

@@ -543,7 +543,7 @@ def recordupdates(repo, action, branchmerge):
                 # of that file some time in the past. Thus our
                 # merge will appear as a normal local file
                 # modification.
-                repo.dirstate.normaldirty(fd)
+                repo.dirstate.normallookup(fd)
                 if move:
                     repo.dirstate.forget(f)
         elif m == "d": # directory rename

tests/test-issue522

@@ -0,0 +1,31 @@
+#!/bin/sh
+
+# In the merge below, the file "foo" has the same contents in both
+# parents, but if we look at the file-level history, we'll notice that
+# the version in p1 is an ancestor of the version in p2.  This test
+# makes sure that we'll use the version from p2 in the manifest of the
+# merge revision.
+
+hg init repo
+cd repo
+
+echo foo > foo
+hg ci -d '0 0' -qAm 'add foo'
+
+echo bar >> foo
+hg ci -d '0 0' -m 'change foo'
+
+hg backout -d '0 0' -r tip -m 'backout changed foo'
+
+hg up -C 0
+touch bar
+hg ci -d '0 0' -qAm 'add bar'
+
+hg merge --debug
+hg debugstate | grep foo
+hg st -A foo
+hg ci -d '0 0' -m 'merge'
+
+hg manifest --debug | grep foo
+hg debugindex .hg/store/data/foo.i
+

tests/test-issue522.out

@@ -0,0 +1,17 @@
+reverting foo
+changeset 2:4d9e78aaceee backs out changeset 1:b515023e500e
+1 files updated, 0 files merged, 0 files removed, 0 files unresolved
+resolving manifests
+ overwrite None partial False
+ ancestor bbd179dfa0a7 local 71766447bdbb+ remote 4d9e78aaceee
+ foo: remote is newer -> g
+getting foo
+1 files updated, 0 files merged, 0 files removed, 0 files unresolved
+(branch merge, don't forget to commit)
+n   0         -2 unset             foo
+M foo
+c6fc755d7e68f49f880599da29f15add41f42f5a 644 foo
+   rev    offset  length   base linkrev nodeid       p1           p2
+     0         0       5      0       0 2ed2a3912a0b 000000000000 000000000000
+     1         5       9      1       1 6f4310b00b9a 2ed2a3912a0b 000000000000
+     2        14       5      2       2 c6fc755d7e68 6f4310b00b9a 000000000000