largefiles: keep largefiles from colliding with normal one during linear merge

Before this patch, linear merging of modified or newly added largefile
causes unexpected result, if (1) largefile collides with same name
normal one in the target revision and (2) "local" largefile is chosen,
even though branch merging between such revisions doesn't.

Expected result of such linear merging is:

  (1) (not yet recorded) largefile is kept in the working directory
  (2) largefile is marked as (re-)"added"
  (3) colliding normal file is marked as "removed"

But actual result is:

  (1) largefile in the working directory is unlinked
  (2) largefile is marked as "normal" (so treated as "missing")
  (3) the dirstate entry for colliding normal file is just dropped

(1) is very serious, because there is no way to restore temporarily
modified largefiles.

(3) prevents the next commit from adding the manifest with correct
"removal of (normal) file" information for newly created changeset.

The root cause of this problem is putting "lfile" into "actions['r']"
in linear-merging case. At liner merging, "actions['r']" causes:

  - unlinking "target file" in the working directory, but "lfile" as
    "target file" is also largefile itself in this case

  - dropping the dirstate entry for target file

"actions['f']" (= "forget") does only the latter, and this is reason
why this patch doesn't choose putting "lfile" into it instead of
"actions['r']".

This patch newly introduces action "lfmr" (LargeFiles: Mark as
Removed) to mark colliding normal file as "removed" without unlinking
it.

This patch uses "hg debugdirstate" instead of "hg status" in test,
because:

  - choosing "local largefile" hides "removed" status of "remote
    normal file" in "hg status" output, and

  - "hg status" for "large2" in this case has another problem fixed in
    the subsequent patch

hgext/largefiles/overrides.py

@@ -430,6 +430,7 b' def overridecalculateupdates(origfn, rep'
     removes = set(a[0] for a in actions['r'])
 
     newglist = []
+    lfmr = [] # LargeFiles: Mark as Removed
     for action in actions['g']:
         f, args, msg = action
         splitstandin = f and lfutil.splitstandin(f)
@@ -456,7 +457,16 b' def overridecalculateupdates(origfn, rep'
                     'keep (l)argefile or use (n)ormal file?'
                     '$$ &Largefile $$ &Normal file') % lfile
             if repo.ui.promptchoice(msg, 0) == 0:
-                actions['r'].append((lfile, None, msg))
+                if branchmerge:
+                    # largefile can be restored from standin safely
+                    actions['r'].append((lfile, None, msg))
+                else:
+                    # "lfile" should be marked as "removed" without
+                    # removal of itself
+                    lfmr.append((lfile, None, msg))
+
+                    # linear-merge should treat this largefile as 're-added'
+                    actions['a'].append((standin, None, msg))
             else:
                 actions['r'].append((standin, None, msg))
                 newglist.append((lfile, (p2.flags(lfile),), msg))
@@ -465,9 +475,22 b' def overridecalculateupdates(origfn, rep'
 
     newglist.sort()
     actions['g'] = newglist
+    if lfmr:
+        lfmr.sort()
+        actions['lfmr'] = lfmr
 
     return actions
 
+def mergerecordupdates(orig, repo, actions, branchmerge):
+    if 'lfmr' in actions:
+        # this should be executed before 'orig', to execute 'remove'
+        # before all other actions
+        for lfile, args, msg in actions['lfmr']:
+            repo.dirstate.remove(lfile)
+
+    return orig(repo, actions, branchmerge)
+
+
 # Override filemerge to prompt the user about how they wish to merge
 # largefiles. This will handle identical edits without prompting the user.
 def overridefilemerge(origfn, repo, mynode, orig, fcd, fco, fca, labels=None):

hgext/largefiles/uisetup.py

@@ -99,6 +99,8 b' def uisetup(ui):'
                                     overrides.overridecheckunknownfile)
     entry = extensions.wrapfunction(merge, 'calculateupdates',
                                     overrides.overridecalculateupdates)
+    entry = extensions.wrapfunction(merge, 'recordupdates',
+                                    overrides.mergerecordupdates)
     entry = extensions.wrapfunction(filemerge, 'filemerge',
                                     overrides.overridefilemerge)
     entry = extensions.wrapfunction(cmdutil, 'copy',

tests/test-largefiles-update.t

@@ -220,4 +220,48 b' Test that linear merge can detect modifi'
   $ cat .hglf/large1
   ba94c2efe5b7c5e0af8d189295ce00553b0612b7
 
+Test a linear merge to a revision containing same-name normal file
+
+  $ hg update -q -C 3
+  $ hg remove large2
+  $ echo 'large2 as normal file' > large2
+  $ hg add large2
+  $ echo 'large3 as normal file' > large3
+  $ hg add large3
+  $ hg commit -m '#5'
+  $ hg manifest
+  .hglf/large1
+  large2
+  large3
+  normal1
+
+(modified largefile is already switched to normal)
+
+  $ hg update -q -C 2
+  $ echo 'modified large2 for linear merge' > large2
+  $ hg update -q 5
+  local changed .hglf/large2 which remote deleted
+  use (c)hanged version or (d)elete? c
+  remote turned local largefile large2 into a normal file
+  keep (l)argefile or use (n)ormal file? l
+  $ hg debugdirstate --nodates | grep large2
+  a   0         -1 .hglf/large2
+  r   0          0 large2
+  $ cat large2
+  modified large2 for linear merge
+
+(added largefile is already committed as normal)
+
+  $ hg update -q -C 2
+  $ echo 'large3 as large file for linear merge' > large3
+  $ hg add --large large3
+  $ hg update -q 5
+  remote turned local largefile large3 into a normal file
+  keep (l)argefile or use (n)ormal file? l
+  $ hg debugdirstate --nodates | grep large3
+  a   0         -1 .hglf/large3
+  r   0          0 large3
+  $ cat large3
+  large3 as large file for linear merge
+
   $ cd ..