Rev 2981: Per-file graph heads detection during commit for pack repositories. in http://people.ubuntu.com/~robertc/baz2.0/commit.merge-speed

[Robert Collins]

At http://people.ubuntu.com/~robertc/baz2.0/commit.merge-speed

------------------------------------------------------------
revno: 2981
revision-id:robertc at robertcollins.net-20071112225758-co7hf9fpdonhkzs8
parent: robertc at robertcollins.net-20071112210301-da1gpusi4p68xhxj
committer: Robert Collins <robertc at robertcollins.net>
branch nick: commit.merge-speed
timestamp: Tue 2007-11-13 09:57:58 +1100
message:
  Per-file graph heads detection during commit for pack repositories.
modified:
  bzrlib/graph.py                graph_walker.py-20070525030359-y852guab65d4wtn0-1
  bzrlib/index.py                index.py-20070712131115-lolkarso50vjr64s-1
  bzrlib/repofmt/pack_repo.py    pack_repo.py-20070813041115-gjv5ma7ktfqwsjgn-1
  bzrlib/repository.py           rev_storage.py-20051111201905-119e9401e46257e3
=== modified file 'bzrlib/graph.py'
--- a/bzrlib/graph.py	2007-10-24 06:48:13 +0000
+++ b/bzrlib/graph.py	2007-11-12 22:57:58 +0000
@@ -19,7 +19,6 @@
     tsort,
     )
 from bzrlib.deprecated_graph import (node_distances, select_farthest)
-from bzrlib.revision import NULL_REVISION
 
 # DIAGRAM of terminology
 #       A

=== modified file 'bzrlib/index.py'
--- a/bzrlib/index.py	2007-10-18 01:17:01 +0000
+++ b/bzrlib/index.py	2007-11-12 22:57:58 +0000
@@ -32,6 +32,7 @@
 lazy_import(globals(), """
 from bzrlib import trace
 from bzrlib.bisect_multi import bisect_multi_bytes
+from bzrlib.revision import NULL_REVISION
 from bzrlib.trace import mutter
 """)
 from bzrlib import debug, errors
@@ -994,6 +995,38 @@
                 self.__class__.__name__,
                 ', '.join(map(repr, self._indices)))
 
+    def get_parents(self, revision_ids):
+        """See StackedParentsProvider.get_parents.
+        
+        This implementation thunks the graph.Graph.get_parents api across to
+        GraphIndex.
+
+        :param revision_ids: An iterable of graph keys for this graph.
+        :return: A list of parent details for each key in revision_ids.
+            Each parent details will be one of:
+             * None when the key was missing
+             * (NULL_REVISION,) when the key has no parents.
+             * (parent_key, parent_key...) otherwise.
+        """
+        search_keys = set(revision_ids)
+        if NULL_REVISION in search_keys:
+            search_keys.remove(NULL_REVISION)
+        found_parents = {NULL_REVISION:[]}
+        for index, key, value, refs in self.iter_entries(search_keys):
+            parents = refs[0]
+            if not parents:
+                parents = (NULL_REVISION,)
+            else:
+                parents = refs[0]
+            found_parents[key] = parents
+        result = []
+        for key in revision_ids:
+            try:
+                result.append(found_parents[key])
+            except KeyError:
+                result.append(None)
+        return result
+
     def insert_index(self, pos, index):
         """Insert a new index in the list of indices to query.
 

=== modified file 'bzrlib/repofmt/pack_repo.py'
--- a/bzrlib/repofmt/pack_repo.py	2007-11-07 13:10:37 +0000
+++ b/bzrlib/repofmt/pack_repo.py	2007-11-12 22:57:58 +0000
@@ -26,6 +26,7 @@
         pack,
         ui,
         )
+from bzrlib.graph import Graph
 from bzrlib.index import (
     GraphIndex,
     GraphIndexBuilder,
@@ -72,11 +73,24 @@
     added text, reducing memory and object pressure.
     """
 
+    def __init__(self, repository, parents, config, timestamp=None,
+                 timezone=None, committer=None, revprops=None,
+                 revision_id=None):
+        CommitBuilder.__init__(self, repository, parents, config,
+            timestamp=timestamp, timezone=timezone, committer=committer,
+            revprops=revprops, revision_id=revision_id)
+        self._file_graph = Graph(
+            repository._pack_collection.text_index.combined_index)
+
     def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
         return self.repository._pack_collection._add_text_to_weave(file_id,
             self._new_revision_id, new_lines, parents, nostore_sha,
             self.random_revid)
 
+    def heads(self, file_id, revision_ids):
+        keys = [(file_id, revision_id) for revision_id in revision_ids]
+        return set([key[1] for key in self._file_graph.heads(keys)])
+
 
 class PackRootCommitBuilder(RootCommitBuilder):
     """A subclass of RootCommitBuilder to add texts with pack semantics.
@@ -85,11 +99,24 @@
     added text, reducing memory and object pressure.
     """
 
+    def __init__(self, repository, parents, config, timestamp=None,
+                 timezone=None, committer=None, revprops=None,
+                 revision_id=None):
+        CommitBuilder.__init__(self, repository, parents, config,
+            timestamp=timestamp, timezone=timezone, committer=committer,
+            revprops=revprops, revision_id=revision_id)
+        self._file_graph = Graph(
+            repository._pack_collection.text_index.combined_index)
+
     def _add_text_to_weave(self, file_id, new_lines, parents, nostore_sha):
         return self.repository._pack_collection._add_text_to_weave(file_id,
             self._new_revision_id, new_lines, parents, nostore_sha,
             self.random_revid)
 
+    def heads(self, file_id, revision_ids):
+        keys = [(file_id, revision_id) for revision_id in revision_ids]
+        return set([key[1] for key in self._file_graph.heads(keys)])
+
 
 class Pack(object):
     """An in memory proxy for a pack and its indices.

=== modified file 'bzrlib/repository.py'
--- a/bzrlib/repository.py	2007-11-12 21:03:01 +0000
+++ b/bzrlib/repository.py	2007-11-12 22:57:58 +0000
@@ -74,8 +74,8 @@
     # the default CommitBuilder does not manage trees whose root is versioned.
     _versioned_root = False
 
-    def __init__(self, repository, parents, config, timestamp=None, 
-                 timezone=None, committer=None, revprops=None, 
+    def __init__(self, repository, parents, config, timestamp=None,
+                 timezone=None, committer=None, revprops=None,
                  revision_id=None):
         """Initiate a CommitBuilder.