Merge lp:~rockstar/loggerhead/revnocache into lp:loggerhead

[Paul Hummer]

[Paul Hummer (rockstar)]

This branch adds a MergeGraphCache from the bzr-revnocache plugin. This
allowed me to remove two of the whole history cache stuff. Something that I
had to do was convert all internal references to a revno to tuples. All the
tests pass now.

As the reviewer, please check that my implementation is sane. I'm sure there
will be MANY comments and changes I need to make, and I welcome them.

 reviewer mwhudson

Cheers,
Paul

[Michael Hudson-Doyle (mwhudson)]

Hm, I think this branch may be guilty of trying to do too many things at once. I think there are two separate tasks: (1) use revnocache to make accessing the merge sorted revision graph more efficient (2) reducing the use of whole-history data in history.py. This sort of tries to do both :/

Also, it makes loggerhead hard-depend on revnocache being installed, which isn't great. So, sorry, I think we'll have to try again here... and I'm going on leave for a week so we don't really have a chance to talk about what would be better. Oops.

review: Needs Resubmitting

Unmerged revisions

338. By Paul Hummer on 2009-04-08

Removed unneeded code

337. By Paul Hummer on 2009-04-08

Fixed the last test

336. By Paul Hummer on 2009-04-08

Only one failing test now

335. By Paul Hummer on 2009-04-08

Fixed a bug in fix_revid

334. By Paul Hummer on 2009-04-08

Added converters for dotted revnos

333. By Paul Hummer on 2009-04-08

Fixed a bug in file views

332. By Paul Hummer on 2009-04-08

Eliminated the need for revgraphcache

331. By Paul Hummer on 2009-04-08

Added call to clear()

330. By Paul Hummer on 2009-04-08

Added hack to regenerate cache

329. By Paul Hummer on 2009-04-08

Removed pdb call

=== modified file 'loggerhead/history.py'
--- loggerhead/history.py	2009-03-19 20:04:24 +0000
+++ loggerhead/history.py	2009-04-08 21:05:38 +0000
@@ -42,11 +42,19 @@
 from loggerhead import util
 from loggerhead.wholehistory import compute_whole_history_data
 
+from bzrlib.plugin import load_plugins
+
+load_plugins()
+
 import bzrlib
 import bzrlib.branch
 import bzrlib.delta
 import bzrlib.diff
 import bzrlib.errors
+from bzrlib.plugins.revnocache import _branch_iter_merge_graph
+from bzrlib.plugins.revnocache.mergegraphcache import MergeGraphCache
+from bzrlib.plugins.revnocache.revnocache import RevnoCache
+from bzrlib.plugins.revnocache.revgraphcache import RevGraphCache
 import bzrlib.progress
 import bzrlib.revision
 import bzrlib.textfile
@@ -177,6 +185,16 @@
                 file_id=file_id))
 
 
+def make_mergegraphcache(branch):
+    '''Make the mergegraphcache data structure for writing to the cache.'''
+    return _branch_iter_merge_graph(branch)
+
+
+def convert_revnostr_to_tuple(revno):
+    '''Convert a revnostr (dotted or not) to its tuple form.'''
+    return tuple([int(i) for i in revno.split('.')])
+
+
 class History (object):
     """Decorate a branch to provide information for rendering.
 
@@ -199,18 +217,40 @@
 
         self.last_revid = branch.last_revision()
 
+        self.mergegraphcache = MergeGraphCache(branch=self._branch)
+        if not self.mergegraphcache.exists():
+            self.mergegraphcache.write(make_mergegraphcache(self._branch))
+        else:
+            revision = self.get_mergegraph_rev(revid=self.last_revid)
+            if not revision:
+                self.mergegraphcache.clear()
+                self.mergegraphcache.write(make_mergegraphcache(self._branch))
+
         whole_history_data = whole_history_data_cache.get(self.last_revid)
         if whole_history_data is None:
             whole_history_data = compute_whole_history_data(branch)
             whole_history_data_cache[self.last_revid] = whole_history_data
 
-        (self._revision_graph, self._full_history, self._revision_info,
-         self._revno_revid, self._merge_sort, self._where_merged,
-         ) = whole_history_data
+        (self._revision_graph, self._full_history, self._merge_sort,
+         self._where_merged) = whole_history_data
 
     def use_file_cache(self, cache):
         self._file_change_cache = cache
 
+    def get_mergegraph_rev(self, revid=None, revno=None):
+        '''Return the merge graph cache record for a given revid.'''
+        if revid:
+            for revision in self.mergegraphcache.iter():
+                if revision[1] == revid:
+                    return revision
+        if revno:
+            if type(revno) == str:
+                revno = convert_revnostr_to_tuple(revno)
+            for revision in self.mergegraphcache.iter():
+                if revision[3] == revno:
+                    return revision
+        return None
+
     @property
     def has_revisions(self):
         return not bzrlib.revision.is_null(self.last_revid)
@@ -219,12 +259,13 @@
         return self._branch.get_config()
 
     def get_revno(self, revid):
-        if revid not in self._revision_info:
+        try:
+            revno_str = '.'.join(
+                [str(i) for i in self.get_mergegraph_rev(revid=revid)[3]])
+            return revno_str
+        except TypeError:
             # ghost parent?
             return 'unknown'
-        (seq, revid, merge_depth,
-         revno_str, end_of_merge) = self._revision_info[revid]
-        return revno_str
 
     def get_revids_from(self, revid_list, start_revid):
         """
@@ -238,7 +279,8 @@
 
         def introduced_revisions(revid):
             r = set([revid])
-            seq, revid, md, revno, end_of_merge = self._revision_info[revid]
+            seq, revid, md, revno, end_of_merge = self.get_mergegraph_rev(
+                revid=revid)
             i = seq + 1
             while i < len(self._merge_sort) and self._merge_sort[i][2] > md:
                 r.add(self._merge_sort[i][1])
@@ -358,11 +400,12 @@
             return revid
         if revid == 'head:':
             return self.last_revid
-        try:
-            if self.revno_re.match(revid):
-                revid = self._revno_revid[revid]
-        except KeyError:
-            raise bzrlib.errors.NoSuchRevision(self._branch_nick, revid)
+        if self.revno_re.match(revid):
+            revno = convert_revnostr_to_tuple(revid)
+            try:
+                return self.get_mergegraph_rev(revno=revno)[1]
+            except TypeError:
+                raise bzrlib.errors.NoSuchRevision(self._branch_nick, revid)
         return revid
 
     def get_file_view(self, revid, file_id):

=== modified file 'loggerhead/wholehistory.py'
--- loggerhead/wholehistory.py	2008-12-05 20:09:14 +0000
+++ loggerhead/wholehistory.py	2009-04-08 06:03:22 +0000
@@ -75,5 +75,4 @@
 
     log.info('built revision graph cache: %r secs' % (time.time() - z))
 
-    return (_revision_graph, _full_history, _revision_info,
-            _revno_revid, _merge_sort, _where_merged)
+    return (_revision_graph, _full_history, _merge_sort, _where_merged)