performance: speedup computation of hidden revisions

In their current state, revset calls can be very costlys, as we
test predicates on the entire repository. The hidden filter is very
widely used, and needs to be very fast.

This change drops revset calls in favor of direct revision manipulation.

Performance test on my Mercurial checkout

- 19857 total changesets,
- 1584 obsolete changesets,
- 13310 obsolescence markers.

Before:
 ! hidden
 ! wall 0.077553

After this changes:
  ! hidden
  ! wall 0.011230

Performance test on a Mozilla central checkout:

- 117293 total changesets,
- 1 obsolete changeset,
- 1 obsolescence marker.

Before:
  ! hidden
  ! wall 0.389472

After:
  ! hidden
  ! wall 0.000079

mercurial/repoview.py

@@ -9,6 +9,7 b''
 import copy
 import phases
 import util
+import obsolete, bookmarks, revset
 
 
 def computehidden(repo):
@@ -16,10 +17,19 b' def computehidden(repo):'
 
     During most operation hidden should be filtered."""
     assert not repo.changelog.filteredrevs
-    if repo.obsstore:
-        ### hide extinct changeset that are not accessible by any mean
-        hiddenquery = 'extinct() - ::(parents() + bookmark())'
-        return frozenset(repo.revs(hiddenquery))
+    hideable = obsolete.getrevs(repo, 'obsolete')
+    if hideable:
+        cl = repo.changelog
+        firsthideable = min(hideable)
+        revs = cl.revs(start=firsthideable)
+        blockers = [r for r in revset._children(repo, revs, hideable)
+                      if r not in hideable]
+        for par in repo[None].parents():
+            blockers.append(par.rev())
+        for bm in bookmarks.listbookmarks(repo).values():
+            blockers.append(repo[bm].rev())
+        blocked = cl.ancestors(blockers, inclusive=True)
+        return frozenset(r for r in hideable if r not in blocked)
     return frozenset()
 
 def computeunserved(repo):